[{"title":"AI Image Generator for Inkscape: Stable Diffusion \u0026 DALL·E","permalink":"https://youvenz.github.io/blog/2026-03-05-ai-image-generator-for-inkscape-stable-diffusion-dalle/","summary":"Generate AI Images Directly Inside Inkscape Using Stable Diffusion \u0026amp; DALL·E Every time you want to test an AI-generated image concept, you\u0026rsquo;re stuck in the same loop: sketch in Inkscape → open browser → navigate to DALL·E or Stable Diffusion → wait for generation → download → import back into Inkscape → repeat. You lose context, waste time, and break creative momentum.\nThe AI Image Generator extension eliminates that friction entirely.\n","content":"Generate AI Images Directly Inside Inkscape Using Stable Diffusion \u0026amp;amp; DALL·E Every time you want to test an AI-generated image concept, you\u0026amp;rsquo;re stuck in the same loop: sketch in Inkscape → open browser → navigate to DALL·E or Stable Diffusion → wait for generation → download → import back into Inkscape → repeat. You lose context, waste time, and break creative momentum.\nThe AI Image Generator extension eliminates that friction entirely.\nWhat This Is The AI Image Generator is a free, …","tags":["Inkscape","Stable Diffusion","DALL-E","AI Image Generation","Flux","Creative Workflow","Open-source Extensions","AI Tools for Designers"],"section":"blog"},{"title":"AI SVG Generator: Create Diagrams in Inkscape with LLMs","permalink":"https://youvenz.github.io/blog/2026-03-05-ai-svg-generator-create-diagrams-in-inkscape-with-llms/","summary":"Generate SVG Diagrams with an LLM Directly Inside Inkscape You\u0026rsquo;ve spent the last hour manually drawing a neural network architecture in Inkscape—boxes, arrows, labels, grouped elements—only to realize your client wants it restructured. You start over. This is the pain point: creating complex technical SVGs by hand is repetitive, time-consuming, and fragile to changes. What if you could describe the diagram in plain English and have a fully editable SVG appear on your canvas in seconds?\n","content":"Generate SVG Diagrams with an LLM Directly Inside Inkscape You\u0026amp;rsquo;ve spent the last hour manually drawing a neural network architecture in Inkscape—boxes, arrows, labels, grouped elements—only to realize your client wants it restructured. You start over. This is the pain point: creating complex technical SVGs by hand is repetitive, time-consuming, and fragile to changes. What if you could describe the diagram in plain English and have a fully editable SVG appear on your canvas in seconds? …","tags":["Inkscape","AI SVG Generator","LLM","Diagram generation","Vector graphics automation","Claude API","Technical illustration","AI-assisted design"],"section":"blog"},{"title":"AI Text Generator for Inkscape: LLM Inside","permalink":"https://youvenz.github.io/blog/2026-03-05-ai-text-generator-for-inkscape-llm-inside/","summary":"Generate Text with an LLM Directly Inside Inkscape — For Designers Who Want AI Copy Without Leaving the App You\u0026rsquo;re three iterations deep on a design mockup. The headline needs tweaking. The body copy should be shorter. And wait—the client just asked for a French version. You close Inkscape, open ChatGPT, paste the text, copy the result, switch back, paste it in, adjust the font, and realize it doesn\u0026rsquo;t fit anymore.\n","content":"Generate Text with an LLM Directly Inside Inkscape — For Designers Who Want AI Copy Without Leaving the App You\u0026amp;rsquo;re three iterations deep on a design mockup. The headline needs tweaking. The body copy should be shorter. And wait—the client just asked for a French version. You close Inkscape, open ChatGPT, paste the text, copy the result, switch back, paste it in, adjust the font, and realize it doesn\u0026amp;rsquo;t fit anymore.\nThere\u0026amp;rsquo;s a better way. What if you could generate, rewrite, …","tags":["Inkscape","AI Text Generator","LLM extension","ChatGPT integration","Claude API","design automation","AI for designers","local LLM"],"section":"blog"},{"title":"Anaconda to Apptainer: Reproducible Python Environments","permalink":"https://youvenz.github.io/blog/2026-03-05-anaconda-to-apptainer-reproducible-python-environments/","summary":"Build Reproducible Python Environments with Anaconda to Apptainer — Without Broken HPC Deployments You\u0026rsquo;ve spent weeks perfecting a Python environment locally. Dependencies are locked in. Your code runs flawlessly on your laptop. Then you push it to the HPC cluster, and everything breaks.\nMissing libraries. Version conflicts. Runtime errors at 3 AM when your job finally reaches the queue. The problem: your local Conda environment doesn\u0026rsquo;t travel. It\u0026rsquo;s fragile, system-dependent, and impossible to reproduce across different machines.\n","content":"Build Reproducible Python Environments with Anaconda to Apptainer — Without Broken HPC Deployments You\u0026amp;rsquo;ve spent weeks perfecting a Python environment locally. Dependencies are locked in. Your code runs flawlessly on your laptop. Then you push it to the HPC cluster, and everything breaks.\nMissing libraries. Version conflicts. Runtime errors at 3 AM when your job finally reaches the queue. The problem: your local Conda environment doesn\u0026amp;rsquo;t travel. It\u0026amp;rsquo;s fragile, system-dependent, …","tags":["Apptainer","Singularity","Conda","Python containers","HPC reproducibility","containerization","environment management","ML deployment"],"section":"blog"},{"title":"Apptainer \u0026 Conda: Reproducible HPC Python Environments","permalink":"https://youvenz.github.io/blog/2026-03-05-apptainer-conda-reproducible-hpc-python-environments/","summary":"Build Reproducible Python Environments with Apptainer \u0026amp; Conda — For HPC Researchers Your Python script runs perfectly on your laptop. You transfer it to the cluster. Runtime fails. Missing dependencies. Version conflicts. Different Python builds. You spend hours debugging environment mismatches instead of running science.\nThis is the researcher\u0026rsquo;s tax: environment reproduction across machines is broken by default.\nWhat This Solves Apptainer (formerly Singularity) is a containerization tool built for HPC clusters. Unlike Docker, it doesn\u0026rsquo;t require root privileges and integrates seamlessly with cluster job schedulers like SLURM and PBS. Combined with Conda, it lets you package a complete, reproducible Python environment—dependencies, versions, and all—into a single .sif file that runs identically on your laptop, your colleague\u0026rsquo;s machine, and any HPC node.\n","content":"Build Reproducible Python Environments with Apptainer \u0026amp;amp; Conda — For HPC Researchers Your Python script runs perfectly on your laptop. You transfer it to the cluster. Runtime fails. Missing dependencies. Version conflicts. Different Python builds. You spend hours debugging environment mismatches instead of running science.\nThis is the researcher\u0026amp;rsquo;s tax: environment reproduction across machines is broken by default.\nWhat This Solves Apptainer (formerly Singularity) is a containerization …","tags":["Apptainer","Conda","HPC","containerization","reproducible research","Python environments","SLURM","cluster computing"],"section":"blog"},{"title":"Apptainer Fundamentals: Building Your First Container","permalink":"https://youvenz.github.io/blog/2026-03-05-apptainer-fundamentals-building-your-first-container/","summary":"Build Your First Apptainer Container Without Docker — A Practical Guide for HPC Researchers You\u0026rsquo;ve got HPC cluster access, but containerizing your workflow feels like a black box. Singularity (now Apptainer) promises reproducibility and portability, but the documentation jumps between concepts, and you\u0026rsquo;re not sure where to start. You don\u0026rsquo;t want to learn Docker first—you just want a working container on your cluster today.\nHere\u0026rsquo;s the good news: Apptainer is simpler than you think, and you don\u0026rsquo;t need Docker to get started.\n","content":"Build Your First Apptainer Container Without Docker — A Practical Guide for HPC Researchers You\u0026amp;rsquo;ve got HPC cluster access, but containerizing your workflow feels like a black box. Singularity (now Apptainer) promises reproducibility and portability, but the documentation jumps between concepts, and you\u0026amp;rsquo;re not sure where to start. You don\u0026amp;rsquo;t want to learn Docker first—you just want a working container on your cluster today.\nHere\u0026amp;rsquo;s the good news: Apptainer is simpler than …","tags":["Apptainer","Singularity","HPC containers","Container building","Linux containerization","SLURM workflows","Reproducible research","Container deployment"],"section":"blog"},{"title":"Batch Certificate Creation with Inkscape \u0026 Next Generator","permalink":"https://youvenz.github.io/blog/2026-03-05-batch-certificate-creation-with-inkscape-next-generator/","summary":"Batch Certificates Without Manual Design Work — Using Inkscape \u0026amp; Next Generator You just finished running a 50-person workshop. Now you need to generate 50 unique certificates—each with a different name, completion date, grade, and sometimes a different badge image. Doing this manually in Inkscape (or worse, Word) takes hours and introduces typos.\nThis doesn\u0026rsquo;t have to be your workflow.\nWith Inkscape\u0026rsquo;s Next Generator extension, you can automate the entire process: design one certificate template, link it to a CSV spreadsheet with your attendee data, and generate 50+ customized PDFs in minutes. Variable names, conditional images, dynamic grade colors—all from a single batch command.\n","content":"Batch Certificates Without Manual Design Work — Using Inkscape \u0026amp;amp; Next Generator You just finished running a 50-person workshop. Now you need to generate 50 unique certificates—each with a different name, completion date, grade, and sometimes a different badge image. Doing this manually in Inkscape (or worse, Word) takes hours and introduces typos.\nThis doesn\u0026amp;rsquo;t have to be your workflow.\nWith Inkscape\u0026amp;rsquo;s Next Generator extension, you can automate the entire process: design one …","tags":["Inkscape","Next Generator","Batch Certificate Generation","CSV Automation","Course Completion Certificates","PDF Generation","Workshop Certificates","Design Automation"],"section":"blog"},{"title":"Build a Fast Academic Website with Hugo","permalink":"https://youvenz.github.io/blog/2026-03-05-build-a-fast-academic-website-with-hugo/","summary":"Build a Fast Academic Website with Hugo — Without Databases or Server Headaches You\u0026rsquo;re a researcher or PhD student who needs a professional online presence. But the moment you look at WordPress, Wix, or Squarespace, you hit the same walls: slow load times, monthly fees, database management nightmares, and endless plugin updates. What if you could build a sleek, lightning-fast academic site in an afternoon using only markdown files?\nWhat This Is Hugo is a static site generator written in Go that turns markdown files into a fully-built website—no databases, no server-side code, no maintenance headaches. Unlike traditional CMS platforms (WordPress, Drupal), Hugo generates your entire site at build time, meaning your pages load in milliseconds and you have zero security vulnerabilities.\n","content":"Build a Fast Academic Website with Hugo — Without Databases or Server Headaches You\u0026amp;rsquo;re a researcher or PhD student who needs a professional online presence. But the moment you look at WordPress, Wix, or Squarespace, you hit the same walls: slow load times, monthly fees, database management nightmares, and endless plugin updates. What if you could build a sleek, lightning-fast academic site in an afternoon using only markdown files?\nWhat This Is Hugo is a static site generator written in Go …","tags":["Hugo","Static Site Generator","Academic Website","GitHub Pages","Netlify","Markdown","Researchers","PhD Students"],"section":"blog"},{"title":"Convert LaTeX to Word with Pandoc: Preserve Equations","permalink":"https://youvenz.github.io/blog/2026-03-05-convert-latex-to-word-with-pandoc-preserve-equations/","summary":"Convert LaTeX to Word WITHOUT Breaking Equations Using Pandoc You\u0026rsquo;ve spent weeks perfecting your LaTeX document—equations are crisp, references are linked, tables are formatted, and the bibliography flows perfectly. Then your advisor asks: \u0026ldquo;Can you send this as a Word file?\u0026rdquo; Your stomach drops. You\u0026rsquo;ve heard the horror stories: equations become unreadable images, references break, tables collapse, and you\u0026rsquo;re left manually reconstructing everything in Word.\nThere\u0026rsquo;s a better way, and it takes under 5 minutes to set up.\n","content":"Convert LaTeX to Word WITHOUT Breaking Equations Using Pandoc You\u0026amp;rsquo;ve spent weeks perfecting your LaTeX document—equations are crisp, references are linked, tables are formatted, and the bibliography flows perfectly. Then your advisor asks: \u0026amp;ldquo;Can you send this as a Word file?\u0026amp;rdquo; Your stomach drops. You\u0026amp;rsquo;ve heard the horror stories: equations become unreadable images, references break, tables collapse, and you\u0026amp;rsquo;re left manually reconstructing everything in Word. …","tags":["Pandoc","LaTeX to Word conversion","Document conversion tools","Research paper formatting","BibTeX citations","Equation preservation","AI/ML researchers"],"section":"blog"},{"title":"Convert Markdown to Jupyter Notebooks with Jupytext","permalink":"https://youvenz.github.io/blog/2026-03-05-convert-markdown-to-jupyter-notebooks-with-jupytext/","summary":"Convert Markdown to Jupyter Notebooks Using Jupytext — For Researchers Who Need Executable Code Your Markdown files are beautiful but frozen. You\u0026rsquo;ve written detailed documentation with embedded code snippets, but they\u0026rsquo;re just text—no execution, no live plots, no way to tweak parameters and see results instantly. Jupytext solves this in one command: it transforms static Markdown into fully executable Jupyter Notebooks while keeping your source file version-control friendly.\nWhat Jupytext Does Jupytext is a lightweight Python library that converts Markdown files containing code blocks into executable Jupyter Notebooks (.ipynb files). It preserves your Markdown text as notebook cells while converting fenced code blocks into executable code cells. The result: interactive, reproducible research documents you can run, modify, and visualize—all from a source file that remains readable as plain Markdown.\n","content":"Convert Markdown to Jupyter Notebooks Using Jupytext — For Researchers Who Need Executable Code Your Markdown files are beautiful but frozen. You\u0026amp;rsquo;ve written detailed documentation with embedded code snippets, but they\u0026amp;rsquo;re just text—no execution, no live plots, no way to tweak parameters and see results instantly. Jupytext solves this in one command: it transforms static Markdown into fully executable Jupyter Notebooks while keeping your source file version-control friendly.\nWhat …","tags":["Jupytext","Jupyter Notebook","Markdown to Jupyter","Python notebooks","Reproducible research","Data science workflows","Jupyter conversion","Interactive documentation"],"section":"blog"},{"title":"Create Beautiful Course Websites with Jupyter Book","permalink":"https://youvenz.github.io/blog/2026-03-05-create-beautiful-course-websites-with-jupyter-book/","summary":"Create Interactive Course Websites with Jupyter Book — For Researchers \u0026amp; Educators Who Need Updatable, Executable Content Your course materials are already outdated. Your readers can\u0026rsquo;t run your code examples. You can\u0026rsquo;t update without rebuilding everything from scratch. Static PDFs and HTML don\u0026rsquo;t cut it anymore—your audience needs to interact with your content, run code in the browser, and access the latest version instantly.\nJupyter Book solves all three problems at once.\n","content":"Create Interactive Course Websites with Jupyter Book — For Researchers \u0026amp;amp; Educators Who Need Updatable, Executable Content Your course materials are already outdated. Your readers can\u0026amp;rsquo;t run your code examples. You can\u0026amp;rsquo;t update without rebuilding everything from scratch. Static PDFs and HTML don\u0026amp;rsquo;t cut it anymore—your audience needs to interact with your content, run code in the browser, and access the latest version instantly.\nJupyter Book solves all three problems at once. …","tags":["Jupyter Book","course websites","interactive content","Markdown","GitHub Pages","executable notebooks","research education","static site generators"],"section":"blog"},{"title":"Create Beautiful LaTeX Tables with Pandas \u0026 Python","permalink":"https://youvenz.github.io/blog/2026-03-05-create-beautiful-latex-tables-with-pandas-python/","summary":"Generate Publication-Ready LaTeX Tables from CSV Using Pandas — For Researchers \u0026amp; Data Scientists You\u0026rsquo;ve spent hours manually formatting a LaTeX table for your research paper. Then your advisor asks you to re-run the analysis with different parameters. Now you\u0026rsquo;re staring at 200 lines of hand-coded \\hline and \u0026amp; delimiters, knowing you\u0026rsquo;ll have to rebuild the entire table from scratch—and probably introduce formatting errors in the process.\nThis is the reproducibility killer that stops science dead in its tracks.\n","content":"Generate Publication-Ready LaTeX Tables from CSV Using Pandas — For Researchers \u0026amp;amp; Data Scientists You\u0026amp;rsquo;ve spent hours manually formatting a LaTeX table for your research paper. Then your advisor asks you to re-run the analysis with different parameters. Now you\u0026amp;rsquo;re staring at 200 lines of hand-coded \\hline and \u0026amp;amp; delimiters, knowing you\u0026amp;rsquo;ll have to rebuild the entire table from scratch—and probably introduce formatting errors in the process.\nThis is the reproducibility …","tags":["Pandas","LaTeX","Python","Data visualization","Research reproducibility","CSV to LaTeX","Data formatting","Scientific computing"],"section":"blog"},{"title":"Create Course Materials in Markdown + Pandoc","permalink":"https://youvenz.github.io/blog/2026-03-05-create-course-materials-in-markdown-pandoc/","summary":"Create Professional Course Materials in Markdown Using Pandoc — A Complete Workflow for Teachers \u0026amp; Professors You\u0026rsquo;re spending hours formatting exercise sets, lab work, and quizzes in Word or Google Docs—adjusting margins, fixing font inconsistencies, regenerating the same content in three different formats. What if you could write once in Markdown and generate polished PDFs, HTML, and more in seconds?\nMarkdown + Pandoc eliminates this friction entirely. You write your exercises, quizzes, and lab work once in plain text, store it in version control, and convert it instantly to publication-ready PDFs and interactive HTML. No more juggling file formats or losing formatting when sharing with colleagues.\n","content":"Create Professional Course Materials in Markdown Using Pandoc — A Complete Workflow for Teachers \u0026amp;amp; Professors You\u0026amp;rsquo;re spending hours formatting exercise sets, lab work, and quizzes in Word or Google Docs—adjusting margins, fixing font inconsistencies, regenerating the same content in three different formats. What if you could write once in Markdown and generate polished PDFs, HTML, and more in seconds?\nMarkdown + Pandoc eliminates this friction entirely. You write your exercises, …","tags":["Pandoc","Markdown","Course Materials","LaTeX","Document Conversion","Teachers","Educational Content","Version Control"],"section":"blog"},{"title":"Create PowerPoint Slides from Markdown with Pandoc","permalink":"https://youvenz.github.io/blog/2026-03-05-create-powerpoint-slides-from-markdown-with-pandoc/","summary":"Create PowerPoint Presentations from Markdown with Pandoc You\u0026rsquo;ve spent the last hour manually formatting slides in PowerPoint—adjusting fonts, copying text, fixing alignment—only to realize you need to make changes across 20 slides. There\u0026rsquo;s a better way.\nPandoc lets you write your entire presentation in plain text, run a single command, and generate a professionally formatted PowerPoint file in seconds. No clicking. No dragging. No wasted time.\nWhy This Matters If you\u0026rsquo;re a developer, researcher, or content creator, you already know the pain: PowerPoint\u0026rsquo;s interface is slow, changes are tedious, and version control is a nightmare. Markdown + Pandoc flips the script. You write in plain text (which is version-control friendly), separate slides with ---, and convert to .pptx instantly. Bold, italic, lists, images, tables, equations, links, emojis—all supported. Edit further in PowerPoint if you need to, or ship the file as-is.\n","content":"Create PowerPoint Presentations from Markdown with Pandoc You\u0026amp;rsquo;ve spent the last hour manually formatting slides in PowerPoint—adjusting fonts, copying text, fixing alignment—only to realize you need to make changes across 20 slides. There\u0026amp;rsquo;s a better way.\nPandoc lets you write your entire presentation in plain text, run a single command, and generate a professionally formatted PowerPoint file in seconds. No clicking. No dragging. No wasted time.\nWhy This Matters If you\u0026amp;rsquo;re a …","tags":["Pandoc","Markdown","PowerPoint automation","presentation tools","command-line tools","researchers","developers","technical writing"],"section":"blog"},{"title":"D2 Inkscape Extension: Code Diagrams Inside Inkscape","permalink":"https://youvenz.github.io/blog/2026-03-05-d2-inkscape-extension-code-diagrams-inside-inkscape/","summary":"Generate Code-Based Diagrams Directly Inside Inkscape Using D2 — For Researchers \u0026amp; Technical Writers You\u0026rsquo;ve spent 20 minutes manually recreating a system architecture diagram in Inkscape because you changed the layout. Or you exported a PNG from a diagram tool, imported it, and now it\u0026rsquo;s locked—you can\u0026rsquo;t edit text or connections without starting over.\nThe D2 Inkscape extension eliminates this friction: write your diagram as code, generate it live inside Inkscape, and keep full editability. No more conversion workflows. No more locked-in exports.\n","content":"Generate Code-Based Diagrams Directly Inside Inkscape Using D2 — For Researchers \u0026amp;amp; Technical Writers You\u0026amp;rsquo;ve spent 20 minutes manually recreating a system architecture diagram in Inkscape because you changed the layout. Or you exported a PNG from a diagram tool, imported it, and now it\u0026amp;rsquo;s locked—you can\u0026amp;rsquo;t edit text or connections without starting over.\nThe D2 Inkscape extension eliminates this friction: write your diagram as code, generate it live inside Inkscape, and keep …","tags":["D2 diagrams","Inkscape extension","technical writing","system architecture","diagram automation","SVG generation","code-based diagramming","researcher tools"],"section":"blog"},{"title":"Deploy Jupyter Book to GitHub Pages FREE","permalink":"https://youvenz.github.io/blog/2026-03-05-deploy-jupyter-book-to-github-pages-free/","summary":"Deploy Jupyter Book to GitHub Pages Using GitHub Actions You\u0026rsquo;ve built your Jupyter Book locally. It looks great on your machine. But right now, it\u0026rsquo;s trapped there—invisible to the world. You need a live URL to share with collaborators, students, or your audience. You need it updated automatically every time you push, and you need it free. That\u0026rsquo;s exactly what we\u0026rsquo;re solving today.\nGitHub Pages + GitHub Actions = automated, free hosting for your Jupyter Book. Every time you push changes, a workflow automatically rebuilds your book and publishes it live. No manual steps. No paid hosting. Just push → build → live.\n","content":"Deploy Jupyter Book to GitHub Pages Using GitHub Actions You\u0026amp;rsquo;ve built your Jupyter Book locally. It looks great on your machine. But right now, it\u0026amp;rsquo;s trapped there—invisible to the world. You need a live URL to share with collaborators, students, or your audience. You need it updated automatically every time you push, and you need it free. That\u0026amp;rsquo;s exactly what we\u0026amp;rsquo;re solving today.\nGitHub Pages + GitHub Actions = automated, free hosting for your Jupyter Book. Every time you …","tags":["Jupyter Book","GitHub Pages","GitHub Actions","CI/CD","documentation hosting","Python deployment","automated publishing"],"section":"blog"},{"title":"Excalidraw AI Text-to-Diagram for Research Workflows","permalink":"https://youvenz.github.io/blog/2026-03-05-excalidraw-ai-text-to-diagram-for-research-workflows/","summary":"Create Research Workflow Diagrams Without Design Skills — Using Excalidraw\u0026rsquo;s AI Text-to-Diagram You\u0026rsquo;ve spent 30 minutes in PowerPoint trying to align boxes and arrows for your research pipeline. Or you\u0026rsquo;ve stared at a blank Canva canvas wondering where to start. The real problem isn\u0026rsquo;t that you can\u0026rsquo;t draw—it\u0026rsquo;s that you\u0026rsquo;re wasting research time on design instead of thinking about your actual work.\nExcalidraw\u0026rsquo;s text-to-diagram feature solves this: describe your workflow in plain language, and AI generates an editable diagram in seconds.\n","content":"Create Research Workflow Diagrams Without Design Skills — Using Excalidraw\u0026amp;rsquo;s AI Text-to-Diagram You\u0026amp;rsquo;ve spent 30 minutes in PowerPoint trying to align boxes and arrows for your research pipeline. Or you\u0026amp;rsquo;ve stared at a blank Canva canvas wondering where to start. The real problem isn\u0026amp;rsquo;t that you can\u0026amp;rsquo;t draw—it\u0026amp;rsquo;s that you\u0026amp;rsquo;re wasting research time on design instead of thinking about your actual work.\nExcalidraw\u0026amp;rsquo;s text-to-diagram feature solves this: …","tags":["Excalidraw","AI text-to-diagram","research workflows","ML pipelines","diagram generation","workflow visualization","AI tools for researchers","free diagramming"],"section":"blog"},{"title":"Hugo + GitHub Pages: Automate Deployment with GitHub Actions","permalink":"https://youvenz.github.io/blog/2026-03-05-hugo-github-pages-automate-deployment-with-github-actions/","summary":"Deploy a Hugo Website to GitHub Pages Using GitHub Actions — Student \u0026amp; Researcher Guide You\u0026rsquo;ve built a beautiful Hugo website locally. It works perfectly when you run hugo server. But now what? You\u0026rsquo;re staring at a folder on your laptop, unsure how to share it with the world—or worse, how to update it without manually rebuilding and uploading files every single time. GitHub Pages + GitHub Actions solves this: every time you push changes to your repository, your site rebuilds and deploys automatically. No manual steps. No confusion.\n","content":"Deploy a Hugo Website to GitHub Pages Using GitHub Actions — Student \u0026amp;amp; Researcher Guide You\u0026amp;rsquo;ve built a beautiful Hugo website locally. It works perfectly when you run hugo server. But now what? You\u0026amp;rsquo;re staring at a folder on your laptop, unsure how to share it with the world—or worse, how to update it without manually rebuilding and uploading files every single time. GitHub Pages + GitHub Actions solves this: every time you push changes to your repository, your site rebuilds and …","tags":["Hugo","GitHub Pages","GitHub Actions","Static Site Generation","CI/CD","Web Deployment","Researchers","Students"],"section":"blog"},{"title":"Inkscape Poster Utils: Auto-Layout Academic Headers","permalink":"https://youvenz.github.io/blog/2026-03-05-inkscape-poster-utils-auto-layout-academic-headers/","summary":"Auto-Layout Academic Poster Headers in Inkscape Using Poster Utils You\u0026rsquo;ve spent hours manually positioning your poster title, author names, institutional affiliations, and conference details in Inkscape—only to realize the spacing is inconsistent, the font hierarchy looks amateur, and you need to rebuild it from scratch for your next conference. Poster Utils eliminates this friction entirely, generating professional poster headers in seconds with full customization control.\nWhat This Is Poster Utils is a free Inkscape extension that automatically generates and layouts academic poster headers. Input your title, authors, institutions, and conference name once—separated by simple delimiters—and the extension creates a professionally formatted header with correct spacing, font hierarchy, and institutional attribution mapping. It supports both Inkscape\u0026rsquo;s native text rendering and LaTeX output for publication-quality typography.\n","content":"Auto-Layout Academic Poster Headers in Inkscape Using Poster Utils You\u0026amp;rsquo;ve spent hours manually positioning your poster title, author names, institutional affiliations, and conference details in Inkscape—only to realize the spacing is inconsistent, the font hierarchy looks amateur, and you need to rebuild it from scratch for your next conference. Poster Utils eliminates this friction entirely, generating professional poster headers in seconds with full customization control.\nWhat This Is …","tags":["Inkscape","Poster Utils","academic posters","auto-layout","extensions","typography","research visualization","academic publishing"],"section":"blog"},{"title":"LaTeX in Inkscape: The Correct Way (Tutorial)","permalink":"https://youvenz.github.io/blog/2026-03-05-latex-in-inkscape-the-correct-way-tutorial/","summary":"Embed LaTeX Equations in Inkscape Without Extensions — A Beginner\u0026rsquo;s Guide to Two Methods You\u0026rsquo;ve designed a technical poster in Inkscape and now need to add a complex equation. You\u0026rsquo;ve heard LaTeX is the way to go, but you\u0026rsquo;re stuck: Do you need to install extensions? Will it break your workflow? Can you actually edit equations after you place them?\nThis is the friction point that stops most beginners from using LaTeX in Inkscape at all.\n","content":"Embed LaTeX Equations in Inkscape Without Extensions — A Beginner\u0026amp;rsquo;s Guide to Two Methods You\u0026amp;rsquo;ve designed a technical poster in Inkscape and now need to add a complex equation. You\u0026amp;rsquo;ve heard LaTeX is the way to go, but you\u0026amp;rsquo;re stuck: Do you need to install extensions? Will it break your workflow? Can you actually edit equations after you place them?\nThis is the friction point that stops most beginners from using LaTeX in Inkscape at all.\nWhat This Is Inkscape has two …","tags":["Inkscape","LaTeX","TexText","PDF-LaTeX","Technical illustration","Vector graphics","Equation editing","Scientific posters"],"section":"blog"},{"title":"LaTeX in VSCode 2026: LaTeX Workshop Complete Setup","permalink":"https://youvenz.github.io/blog/2026-03-05-latex-in-vscode-2026-latex-workshop-complete-setup/","summary":"Set Up LaTeX in VSCode Without Terminal Headaches — For Researchers \u0026amp; Students Writing Theses You\u0026rsquo;ve started a thesis, research paper, or technical document. You open VSCode—your favorite editor—but LaTeX won\u0026rsquo;t compile. You\u0026rsquo;re stuck toggling between a terminal window, a PDF viewer, and your editor. The setup feels fragmented, slow, and error-prone.\nWhat if you could write, compile, and preview your LaTeX document all in one place?\nThat\u0026rsquo;s what LaTeX Workshop does. This guide gets you from zero to a working LaTeX environment in 15 minutes—no terminal wrestling required.\n","content":"Set Up LaTeX in VSCode Without Terminal Headaches — For Researchers \u0026amp;amp; Students Writing Theses You\u0026amp;rsquo;ve started a thesis, research paper, or technical document. You open VSCode—your favorite editor—but LaTeX won\u0026amp;rsquo;t compile. You\u0026amp;rsquo;re stuck toggling between a terminal window, a PDF viewer, and your editor. The setup feels fragmented, slow, and error-prone.\nWhat if you could write, compile, and preview your LaTeX document all in one place?\nThat\u0026amp;rsquo;s what LaTeX Workshop does. This …","tags":["LaTeX Workshop","VSCode","LaTeX","PDF preview","thesis writing","research tools","technical documentation","IDE setup"],"section":"blog"},{"title":"Load References: Import Bibliography Files into Inkscape","permalink":"https://youvenz.github.io/blog/2026-03-05-load-references-import-bibliography-files-into-inkscape/","summary":"Import Bibliography Files into Inkscape (.bib .ris .enw .json) — Without Manual Copy-Pasting You\u0026rsquo;re designing a research poster or academic publication layout in Inkscape, and you need to add 20+ citations. Right now, you\u0026rsquo;re manually copying references from your .bib file, pasting them into text boxes, reformatting each one, and praying you don\u0026rsquo;t have to update them later. The Load References extension eliminates that friction entirely—your bibliography file becomes a live, editable source inside Inkscape.\n","content":"Import Bibliography Files into Inkscape (.bib .ris .enw .json) — Without Manual Copy-Pasting You\u0026amp;rsquo;re designing a research poster or academic publication layout in Inkscape, and you need to add 20+ citations. Right now, you\u0026amp;rsquo;re manually copying references from your .bib file, pasting them into text boxes, reformatting each one, and praying you don\u0026amp;rsquo;t have to update them later. The Load References extension eliminates that friction entirely—your bibliography file becomes a live, …","tags":["Inkscape","Load References extension","BibTeX","Zotero integration","Bibliography management","Citation formatting","Academic design","Research posters"],"section":"blog"},{"title":"Master Apptainer in 25 Minutes: Build Reproducible Python Environments","permalink":"https://youvenz.github.io/blog/2026-03-05-master-apptainer-in-25-minutes-build-reproducible-python-env/","summary":"Build Reproducible Python Environments with Apptainer (Singularity) — For Researchers Tired of \u0026ldquo;It Works on My Machine\u0026rdquo; Your Python script runs flawlessly on your laptop. Your colleague runs it on theirs and gets import errors. You submit it to the HPC cluster and it crashes with missing CUDA libraries. You spend three hours debugging version conflicts instead of doing research.\nThis is dependency hell, and it\u0026rsquo;s the silent killer of reproducible science.\n","content":"Build Reproducible Python Environments with Apptainer (Singularity) — For Researchers Tired of \u0026amp;ldquo;It Works on My Machine\u0026amp;rdquo; Your Python script runs flawlessly on your laptop. Your colleague runs it on theirs and gets import errors. You submit it to the HPC cluster and it crashes with missing CUDA libraries. You spend three hours debugging version conflicts instead of doing research.\nThis is dependency hell, and it\u0026amp;rsquo;s the silent killer of reproducible science.\nThe good news? A single …","tags":["Apptainer","Singularity","containerization","conda environments","HPC","reproducible science","Python packaging","scientific computing"],"section":"blog"},{"title":"Master Markdown for Research — Write Once, Export Anywhere","permalink":"https://youvenz.github.io/blog/2026-03-05-master-markdown-for-research-write-once-export-anywhere/","summary":"Master Markdown for Research — Write Once, Export Anywhere You\u0026rsquo;re switching between Microsoft Word, Google Docs, and LaTeX for different research outputs. Each tool has its own quirks. You spend 20 minutes reformatting a heading. You copy-paste tables and watch them break. You want to write once and stop fighting with software.\nMarkdown solves this. It\u0026rsquo;s the format that works everywhere—GitHub, LLMs, Jupyter, Obsidian, and Pandoc. And you can learn it in under one hour.\n","content":"Master Markdown for Research — Write Once, Export Anywhere You\u0026amp;rsquo;re switching between Microsoft Word, Google Docs, and LaTeX for different research outputs. Each tool has its own quirks. You spend 20 minutes reformatting a heading. You copy-paste tables and watch them break. You want to write once and stop fighting with software.\nMarkdown solves this. It\u0026amp;rsquo;s the format that works everywhere—GitHub, LLMs, Jupyter, Obsidian, and Pandoc. And you can learn it in under one hour.\nWhat Markdown …","tags":["Markdown","Pandoc","Research writing","Document conversion","Obsidian","LaTeX","Academic workflow"],"section":"blog"},{"title":"Master Text, Connectors \u0026 Layers in Inkscape","permalink":"https://youvenz.github.io/blog/2026-03-05-master-text-connectors-layers-in-inkscape/","summary":"Master Text, Connectors \u0026amp; Layers in Inkscape — For Scientists Building Complex Diagrams You\u0026rsquo;ve drawn shapes in Inkscape. Now you\u0026rsquo;re staring at a half-finished diagram with 15 unlabeled elements, misaligned boxes, and connectors that break when you move things. You need a system to organize this chaos—and you need it fast.\nThat\u0026rsquo;s what this tutorial solves.\nWhat This Is This is Part 2 of the Inkscape for Scientists series. (If you haven\u0026rsquo;t read Part 1 yet, start there—it covers the basics of drawing shapes and working with fills and strokes.)\n","content":"Master Text, Connectors \u0026amp;amp; Layers in Inkscape — For Scientists Building Complex Diagrams You\u0026amp;rsquo;ve drawn shapes in Inkscape. Now you\u0026amp;rsquo;re staring at a half-finished diagram with 15 unlabeled elements, misaligned boxes, and connectors that break when you move things. You need a system to organize this chaos—and you need it fast.\nThat\u0026amp;rsquo;s what this tutorial solves.\nWhat This Is This is Part 2 of the Inkscape for Scientists series. (If you haven\u0026amp;rsquo;t read Part 1 yet, start there—it …","tags":["Inkscape","Scientific Diagrams","Text Formatting","Connector Tool","Layers Panel","Alignment Tools","Vector Graphics","Diagram Design"],"section":"blog"},{"title":"Matplotlib Animation Tutorial: Animate Scientific Data","permalink":"https://youvenz.github.io/blog/2026-03-05-matplotlib-animation-tutorial-animate-scientific-data/","summary":"Animate Scientific Data in Matplotlib — A Step-by-Step Guide for Researchers Creating Dynamic Visualizations You\u0026rsquo;ve spent weeks collecting experimental data. Your results are solid. But when you present them in a static graph, the audience barely glances at it.\nThe real problem isn\u0026rsquo;t your data—it\u0026rsquo;s that a single frame can\u0026rsquo;t show change over time. You need animation. But Matplotlib\u0026rsquo;s FuncAnimation class feels intimidating, and most tutorials skip the crucial structural details that make it actually work.\n","content":"Animate Scientific Data in Matplotlib — A Step-by-Step Guide for Researchers Creating Dynamic Visualizations You\u0026amp;rsquo;ve spent weeks collecting experimental data. Your results are solid. But when you present them in a static graph, the audience barely glances at it.\nThe real problem isn\u0026amp;rsquo;t your data—it\u0026amp;rsquo;s that a single frame can\u0026amp;rsquo;t show change over time. You need animation. But Matplotlib\u0026amp;rsquo;s FuncAnimation class feels intimidating, and most tutorials skip the crucial …","tags":["Matplotlib FuncAnimation","Scientific Visualization Python","Data Animation Tutorial","Research Visualization","Python Animation","Dynamic Plots","Matplotlib Scatter Animation","Scientific Computing"],"section":"blog"},{"title":"Matplotlib Figure Generator: Direct Inkscape Extension","permalink":"https://youvenz.github.io/blog/2026-03-05-matplotlib-figure-generator-direct-inkscape-extension/","summary":"Generate Matplotlib Figures Inside Inkscape — Without Leaving Your Design Tool You\u0026rsquo;ve spent 20 minutes perfecting a bar chart in Matplotlib. Now you need to drop it into your Inkscape poster—but the moment you export as PNG, it\u0026rsquo;s locked. You can\u0026rsquo;t edit the colors. You can\u0026rsquo;t move the legend. You can\u0026rsquo;t adjust the title without regenerating the whole thing in Python, exporting again, and re-importing.\nThere\u0026rsquo;s a better way. The Matplotlib Figure Generator extension lets you build and edit Matplotlib visualizations directly inside Inkscape, with full vector editability and zero export-import friction.\n","content":"Generate Matplotlib Figures Inside Inkscape — Without Leaving Your Design Tool You\u0026amp;rsquo;ve spent 20 minutes perfecting a bar chart in Matplotlib. Now you need to drop it into your Inkscape poster—but the moment you export as PNG, it\u0026amp;rsquo;s locked. You can\u0026amp;rsquo;t edit the colors. You can\u0026amp;rsquo;t move the legend. You can\u0026amp;rsquo;t adjust the title without regenerating the whole thing in Python, exporting again, and re-importing.\nThere\u0026amp;rsquo;s a better way. The Matplotlib Figure Generator …","tags":["Matplotlib","Inkscape","Vector Graphics","Data Visualization","Python Extensions","SVG Generation","Design Automation","Scientific Visualization"],"section":"blog"},{"title":"Matplotlib xkcd Sketch Plots: Hand-Drawn Python Guide","permalink":"https://youvenz.github.io/blog/2026-03-05-matplotlib-xkcd-sketch-plots-hand-drawn-python-guide/","summary":"Transform Your Matplotlib Plots Into Hand-Drawn Sketches Using xkcd Your research plots look crisp and professional—but they also look generic. When presenting findings to a room full of people, a standard line chart disappears into the visual noise. You need something that stops the eye and builds rapport, but you can\u0026rsquo;t sacrifice clarity or credibility. Hand-drawn sketch-style plots solve this: they\u0026rsquo;re engaging, memorable, and still scientifically sound. And they take literally one line of code.\n","content":"Transform Your Matplotlib Plots Into Hand-Drawn Sketches Using xkcd Your research plots look crisp and professional—but they also look generic. When presenting findings to a room full of people, a standard line chart disappears into the visual noise. You need something that stops the eye and builds rapport, but you can\u0026amp;rsquo;t sacrifice clarity or credibility. Hand-drawn sketch-style plots solve this: they\u0026amp;rsquo;re engaging, memorable, and still scientifically sound. And they take literally one …","tags":["Matplotlib","xkcd","Python visualization","research presentation","data visualization","scientific plotting","AI/ML research","data science"],"section":"blog"},{"title":"Mermaid Diagrams in Inkscape: Native Extension Setup","permalink":"https://youvenz.github.io/blog/2026-03-05-mermaid-diagrams-in-inkscape-native-extension-setup/","summary":"Generate Mermaid Diagrams Without Leaving Inkscape — For Academic \u0026amp; Technical Illustrators You\u0026rsquo;re mid-design in Inkscape, and you need to add a flowchart, sequence diagram, or state machine. Right now, your workflow looks like this: switch to a browser, open Mermaid\u0026rsquo;s online editor, create the diagram, export it as PNG or SVG, come back to Inkscape, import it, and hope the formatting survived the conversion. By the time you\u0026rsquo;ve done this three times, you\u0026rsquo;ve lost 20 minutes and broken your creative flow.\n","content":"Generate Mermaid Diagrams Without Leaving Inkscape — For Academic \u0026amp;amp; Technical Illustrators You\u0026amp;rsquo;re mid-design in Inkscape, and you need to add a flowchart, sequence diagram, or state machine. Right now, your workflow looks like this: switch to a browser, open Mermaid\u0026amp;rsquo;s online editor, create the diagram, export it as PNG or SVG, come back to Inkscape, import it, and hope the formatting survived the conversion. By the time you\u0026amp;rsquo;ve done this three times, you\u0026amp;rsquo;ve lost 20 …","tags":["Inkscape","Mermaid","diagram generation","technical illustration","workflow automation","SVG editing","academic writing","extension setup"],"section":"blog"},{"title":"OpenAI API Structured Outputs: Extract Paper Metadata Fast","permalink":"https://youvenz.github.io/blog/2026-03-05-openai-api-structured-outputs-extract-paper-metadata-fast/","summary":"Extract Research Paper Metadata Using OpenAI\u0026rsquo;s Structured Outputs You\u0026rsquo;re three weeks into a systematic literature review. You\u0026rsquo;ve found 200 relevant papers. Now comes the part that makes researchers lose sleep: manually extracting authors, publication year, methodology, key findings, and DOI from each one—copying, pasting, reformatting, praying the data stays consistent.\nWhat if you could automate that entire workflow and have clean, validated JSON output in hours instead of weeks?\nWhat This Is OpenAI\u0026rsquo;s structured outputs force the API to return data in a strict JSON schema you define using Pydantic models. Instead of wrestling with prompt engineering to get the LLM to \u0026ldquo;please format as JSON,\u0026rdquo; you define exactly what fields you want (title, authors, DOI, methodology, etc.), and the API guarantees consistent, validated output every time.\n","content":"Extract Research Paper Metadata Using OpenAI\u0026amp;rsquo;s Structured Outputs You\u0026amp;rsquo;re three weeks into a systematic literature review. You\u0026amp;rsquo;ve found 200 relevant papers. Now comes the part that makes researchers lose sleep: manually extracting authors, publication year, methodology, key findings, and DOI from each one—copying, pasting, reformatting, praying the data stays consistent.\nWhat if you could automate that entire workflow and have clean, validated JSON output in hours instead of …","tags":["OpenAI API","Structured Outputs","Pydantic","Research Automation","Literature Review","Data Extraction","JSON Schema","AI for Research"],"section":"blog"},{"title":"Run LLMs Locally with Llamafile: No Setup Required","permalink":"https://youvenz.github.io/blog/2026-03-05-run-llms-locally-with-llamafile-no-setup-required/","summary":"Run Any LLM Locally Without Setup Using Llamafile You\u0026rsquo;ve tried running local LLMs before. You downloaded dependencies, fought with CUDA versions, debugged GGUF compatibility issues, and waited hours for everything to compile. Then you got a segfault.\nLlamafile changes that. A single executable file runs a full LLM with an OpenAI-compatible API server—no installation, no configuration, no pain.\nWhat Llamafile Actually Is Llamafile packages LLMs into single-file executables using LlamaCPP (a C/C++ inference engine for GGUF models). Download one file, run it, and you get:\n","content":"Run Any LLM Locally Without Setup Using Llamafile You\u0026amp;rsquo;ve tried running local LLMs before. You downloaded dependencies, fought with CUDA versions, debugged GGUF compatibility issues, and waited hours for everything to compile. Then you got a segfault.\nLlamafile changes that. A single executable file runs a full LLM with an OpenAI-compatible API server—no installation, no configuration, no pain.\nWhat Llamafile Actually Is Llamafile packages LLMs into single-file executables using LlamaCPP (a …","tags":["Llamafile","LlamaCPP","Local LLM inference","OpenAI API compatible","GGUF models","GPU acceleration","LLM deployment","AI researchers"],"section":"blog"},{"title":"Run Multiple Jobs with SLURM Array: The Right Way","permalink":"https://youvenz.github.io/blog/2026-03-05-run-multiple-jobs-with-slurm-array-the-right-way/","summary":"Run Multiple Job Variations with SLURM Array Jobs — Without Manual Code Duplication You\u0026rsquo;ve written a solid experiment script. Now you need to run it 50 times with different parameters—learning rates, batch sizes, model architectures—across your HPC cluster.\nThe tempting shortcut: copy-paste your job script 50 times, manually edit each one, submit them all.\nThe reality: you\u0026rsquo;ll waste hours managing parameter files, lose track of which job used which settings, and have no way to scale this to 500 experiments.\n","content":"Run Multiple Job Variations with SLURM Array Jobs — Without Manual Code Duplication You\u0026amp;rsquo;ve written a solid experiment script. Now you need to run it 50 times with different parameters—learning rates, batch sizes, model architectures—across your HPC cluster.\nThe tempting shortcut: copy-paste your job script 50 times, manually edit each one, submit them all.\nThe reality: you\u0026amp;rsquo;ll waste hours managing parameter files, lose track of which job used which settings, and have no way to scale …","tags":["SLURM array jobs","HPC cluster computing","parameter sweeps","job scheduling","reproducible ML workflows","experiment automation","batch processing","cluster management"],"section":"blog"},{"title":"Singularity/Apptainer: Local to Cluster Workflow","permalink":"https://youvenz.github.io/blog/2026-03-05-singularityapptainer-local-to-cluster-workflow/","summary":"Pull, Build, and Deploy Singularity Containers from Local Machine to HPC Clusters You\u0026rsquo;ve got terabytes of data on your HPC cluster and a Python script that works on your laptop. But moving it to the cluster means wrestling with dependency conflicts, mysterious ModuleNotFoundError messages, and copying massive datasets you don\u0026rsquo;t want to duplicate. You need a way to package your entire development environment—dependencies, libraries, Python versions—and run it anywhere without breaking it. That\u0026rsquo;s where Singularity/Apptainer comes in.\n","content":"Pull, Build, and Deploy Singularity Containers from Local Machine to HPC Clusters You\u0026amp;rsquo;ve got terabytes of data on your HPC cluster and a Python script that works on your laptop. But moving it to the cluster means wrestling with dependency conflicts, mysterious ModuleNotFoundError messages, and copying massive datasets you don\u0026amp;rsquo;t want to duplicate. You need a way to package your entire development environment—dependencies, libraries, Python versions—and run it anywhere without breaking …","tags":["Singularity","Apptainer","HPC Clusters","Container Workflow","Docker Alternatives","Linux Containers","Development Environment","Reproducible Research"],"section":"blog"},{"title":"SLURM Job Scheduling: Submit \u0026 Monitor HPC Jobs","permalink":"https://youvenz.github.io/blog/2026-03-05-slurm-job-scheduling-submit-monitor-hpc-jobs/","summary":"Submit \u0026amp; Monitor HPC Jobs Using SLURM — Without Losing Your Data to Path Errors You\u0026rsquo;ve just logged into your university\u0026rsquo;s HPC cluster for the first time. You write what you think is a perfect job script, submit it with sbatch, and 10 minutes later you find a cryptic error in the log file: \u0026ldquo;file does not exist.\u0026rdquo; Your absolute path was wrong. Or you forgot to specify GPU resources and your job waited in the queue for 3 hours doing nothing.\n","content":"Submit \u0026amp;amp; Monitor HPC Jobs Using SLURM — Without Losing Your Data to Path Errors You\u0026amp;rsquo;ve just logged into your university\u0026amp;rsquo;s HPC cluster for the first time. You write what you think is a perfect job script, submit it with sbatch, and 10 minutes later you find a cryptic error in the log file: \u0026amp;ldquo;file does not exist.\u0026amp;rdquo; Your absolute path was wrong. Or you forgot to specify GPU resources and your job waited in the queue for 3 hours doing nothing.\nThis is the moment most …","tags":["SLURM","HPC job scheduling","batch scripts","resource management","sbatch","GPU computing","cluster computing"],"section":"blog"},{"title":"SSH \u0026 SLURM: Transfer Code to HPC Clusters","permalink":"https://youvenz.github.io/blog/2026-03-05-ssh-slurm-transfer-code-to-hpc-clusters/","summary":"Transfer Code to HPC Clusters Using SSH, Singularity/Apptainer \u0026amp; SLURM You\u0026rsquo;ve written working code on your laptop. Now you need to run it on your institution\u0026rsquo;s HPC cluster with 100+ GPUs—but you\u0026rsquo;re staring at a terminal with no idea how to get your files there, containerize your environment, or submit a job without breaking it.\nThis is the moment most researchers feel lost.\nThe HPC Stack in 90 Seconds High-performance computing (HPC) clusters are shared servers with massive CPU and GPU resources. To use them safely and reproducibly, you need three tools working together:\n","content":"Transfer Code to HPC Clusters Using SSH, Singularity/Apptainer \u0026amp;amp; SLURM You\u0026amp;rsquo;ve written working code on your laptop. Now you need to run it on your institution\u0026amp;rsquo;s HPC cluster with 100+ GPUs—but you\u0026amp;rsquo;re staring at a terminal with no idea how to get your files there, containerize your environment, or submit a job without breaking it.\nThis is the moment most researchers feel lost.\nThe HPC Stack in 90 Seconds High-performance computing (HPC) clusters are shared servers with massive …","tags":["SSH","SLURM","Singularity/Apptainer","HPC clusters","GPU computing","data transfer","containerization","rsync"],"section":"blog"},{"title":"Supercharge TeXstudio: Local AI Chat Without APIs","permalink":"https://youvenz.github.io/blog/2026-03-05-supercharge-texstudio-local-ai-chat-without-apis/","summary":"Set Up a Local LLM Inside TeXstudio Without Cloud APIs — For LaTeX Writers Who Want Privacy You\u0026rsquo;re writing a LaTeX paper, and you want AI assistance—but you don\u0026rsquo;t want to pay per API call, send drafts to external servers, or depend on internet connectivity.\nRight now, TeXstudio\u0026rsquo;s AI Chat Assistant only connects to OpenAI or Mistral. There\u0026rsquo;s a third way: run an LLM locally and connect it directly to TeXstudio in 15 minutes.\n","content":"Set Up a Local LLM Inside TeXstudio Without Cloud APIs — For LaTeX Writers Who Want Privacy You\u0026amp;rsquo;re writing a LaTeX paper, and you want AI assistance—but you don\u0026amp;rsquo;t want to pay per API call, send drafts to external servers, or depend on internet connectivity.\nRight now, TeXstudio\u0026amp;rsquo;s AI Chat Assistant only connects to OpenAI or Mistral. There\u0026amp;rsquo;s a third way: run an LLM locally and connect it directly to TeXstudio in 15 minutes.\nWhat This Is TeXstudio 4.8+ includes an AI Chat …","tags":["TeXstudio","Llamafile","Local LLM","LaTeX","OpenAI alternatives","Mistral","Privacy-first AI","AI-assisted writing"],"section":"blog"},{"title":"TexText: LaTeX + Inkscape Integration for Vector Graphics","permalink":"https://youvenz.github.io/blog/2026-03-05-textext-latex-inkscape-integration-for-vector-graphics/","summary":"Combine Inkscape + LaTeX for Stunning Visuals Using TexText You\u0026rsquo;ve spent hours perfecting an equation in LaTeX, then opened Inkscape to add it to a figure, only to realize you need to recompile, export as PDF, and start over. Or worse: your advisor asks you to change a coefficient in a figure, and you\u0026rsquo;re hunting through old source files.\nThe real pain is this: LaTeX gives you typesetting perfection but locks you into a document. Inkscape gives you design freedom but can\u0026rsquo;t handle equations or TikZ code natively. Switching between them kills your workflow and forces endless recompilation cycles.\n","content":"Combine Inkscape + LaTeX for Stunning Visuals Using TexText You\u0026amp;rsquo;ve spent hours perfecting an equation in LaTeX, then opened Inkscape to add it to a figure, only to realize you need to recompile, export as PDF, and start over. Or worse: your advisor asks you to change a coefficient in a figure, and you\u0026amp;rsquo;re hunting through old source files.\nThe real pain is this: LaTeX gives you typesetting perfection but locks you into a document. Inkscape gives you design freedom but can\u0026amp;rsquo;t …","tags":["TexText","Inkscape","LaTeX","TikZ","PGF","vector graphics","research visualization","scientific illustration"],"section":"blog"},{"title":"VS Code Collab Extension: GPU Training in Jupyter","permalink":"https://youvenz.github.io/blog/2026-03-05-vs-code-collab-extension-gpu-training-in-jupyter/","summary":"Run Python Code with GPU Inside VS Code — Without Leaving Your Jupyter Notebooks You\u0026rsquo;ve built a machine learning model locally, but your GPU is sitting idle because you\u0026rsquo;re switching between VS Code and Google Colab every time you need accelerated compute. Or worse—you\u0026rsquo;re uploading files, waiting for notebooks to load, and losing your development flow. The VS Code Collab extension eliminates that friction: connect to CPU, GPU, or TPU compute directly from your Jupyter notebooks without ever leaving your editor.\n","content":"Run Python Code with GPU Inside VS Code — Without Leaving Your Jupyter Notebooks You\u0026amp;rsquo;ve built a machine learning model locally, but your GPU is sitting idle because you\u0026amp;rsquo;re switching between VS Code and Google Colab every time you need accelerated compute. Or worse—you\u0026amp;rsquo;re uploading files, waiting for notebooks to load, and losing your development flow. The VS Code Collab extension eliminates that friction: connect to CPU, GPU, or TPU compute directly from your Jupyter notebooks …","tags":["VS Code Collab","GPU training","Jupyter Notebooks","PyTorch","Local ML development","Google Colab integration","VS Code extensions","ML researchers"],"section":"blog"},{"title":"AI Agents Explained: LLM + Tools + Memory Loop","permalink":"https://youvenz.github.io/blog/2026-03-04-ai-agents-explained-llm-tools-memory-loop/","summary":"AI Agents Explained — Why Your LLM Isn\u0026rsquo;t Actually \u0026ldquo;Doing\u0026rdquo; Anything (Yet) You\u0026rsquo;ve probably used ChatGPT to draft an email or Claude to summarize a paper. You ask, it answers. Simple, right? But here\u0026rsquo;s what most people miss: that\u0026rsquo;s not an agent—that\u0026rsquo;s just a chatbot.\nThe misconception I see constantly in research circles: people think \u0026ldquo;AI agent\u0026rdquo; is just marketing speak for \u0026ldquo;a really good LLM.\u0026rdquo; The reality? An agent is an architecture, not a model. It\u0026rsquo;s the difference between a brain in a jar and a brain connected to hands, eyes, and a notebook.\n","content":"AI Agents Explained — Why Your LLM Isn\u0026amp;rsquo;t Actually \u0026amp;ldquo;Doing\u0026amp;rdquo; Anything (Yet) You\u0026amp;rsquo;ve probably used ChatGPT to draft an email or Claude to summarize a paper. You ask, it answers. Simple, right? But here\u0026amp;rsquo;s what most people miss: that\u0026amp;rsquo;s not an agent—that\u0026amp;rsquo;s just a chatbot.\nThe misconception I see constantly in research circles: people think \u0026amp;ldquo;AI agent\u0026amp;rdquo; is just marketing speak for \u0026amp;ldquo;a really good LLM.\u0026amp;rdquo; The reality? An agent is an …","tags":["AI agents","LLM architecture","agentic workflows","autonomous AI systems","AI tool use","prompt engineering","agent design patterns"],"section":"blog"},{"title":"Anaconda Complete Guide for Beginners | Python Environments","permalink":"https://youvenz.github.io/blog/2026-03-04-anaconda-complete-guide-for-beginners-python-environments/","summary":"Install and Manage Python Environments Using Anaconda — A Beginner\u0026rsquo;s Complete Guide You\u0026rsquo;ve installed Python three different ways, broken your system packages twice, and you still can\u0026rsquo;t figure out why your NumPy version conflicts with your colleague\u0026rsquo;s code. Meanwhile, every tutorial assumes you already know what a \u0026ldquo;virtual environment\u0026rdquo; is—and nobody\u0026rsquo;s explaining why you need one or how to actually use it without breaking everything again.\nThat ends today. This guide walks you through Anaconda from installation to sharing reproducible environments with your team—no prior knowledge required.\n","content":"Install and Manage Python Environments Using Anaconda — A Beginner\u0026amp;rsquo;s Complete Guide You\u0026amp;rsquo;ve installed Python three different ways, broken your system packages twice, and you still can\u0026amp;rsquo;t figure out why your NumPy version conflicts with your colleague\u0026amp;rsquo;s code. Meanwhile, every tutorial assumes you already know what a \u0026amp;ldquo;virtual environment\u0026amp;rdquo; is—and nobody\u0026amp;rsquo;s explaining why you need one or how to actually use it without breaking everything again.\nThat ends today. …","tags":["Anaconda","Python environments","conda package manager","virtual environments","Python dependency management","data science setup","reproducible workflows","ML environment management"],"section":"blog"},{"title":"Inkscape for Scientific Figures: Vector Graphics for Researchers","permalink":"https://youvenz.github.io/blog/2026-03-04-inkscape-for-scientific-figures-vector-graphics-for-research/","summary":"You\u0026rsquo;ve just received reviewer comments on your manuscript: \u0026ldquo;Figure 2 is pixelated and illegible at publication scale.\u0026rdquo; Your carefully prepared PNG screenshots look fine on your screen, but journal editors are flagging them for poor quality. Meanwhile, you\u0026rsquo;re watching colleagues create crisp, professional vector figures—but you don\u0026rsquo;t know where to start, and Adobe Illustrator costs $600/year.\nInkscape is free, open-source vector graphics software that lets you create figures maintaining perfect clarity at any zoom level—exactly what journals require. Unlike raster formats (PNG, JPG) with fixed pixel counts, vector figures scale infinitely without quality loss. This tutorial covers the core tools you need to start creating publication-ready diagrams in 30 minutes.\n","content":"You\u0026amp;rsquo;ve just received reviewer comments on your manuscript: \u0026amp;ldquo;Figure 2 is pixelated and illegible at publication scale.\u0026amp;rdquo; Your carefully prepared PNG screenshots look fine on your screen, but journal editors are flagging them for poor quality. Meanwhile, you\u0026amp;rsquo;re watching colleagues create crisp, professional vector figures—but you don\u0026amp;rsquo;t know where to start, and Adobe Illustrator costs $600/year.\nInkscape is free, open-source vector graphics software that lets you create …","tags":["Inkscape","vector graphics","scientific figures","open-source design","publication quality","research visualization","diagram creation","academic publishing"],"section":"blog"},{"title":"Make Matplotlib Figures Publication Quality","permalink":"https://youvenz.github.io/blog/2026-03-04-make-matplotlib-figures-publication-quality/","summary":"Your Matplotlib Plots Are Getting Your Papers Rejected — Here\u0026rsquo;s the Fix You\u0026rsquo;ve spent weeks perfecting your research, written a solid manuscript, and submitted to your target journal. Then the rejection email arrives: \u0026ldquo;Figures do not meet publication standards.\u0026rdquo;\nThe problem isn\u0026rsquo;t your science—it\u0026rsquo;s your pixelated plots, inconsistent fonts, and low-resolution JPEGs. Journal editors see hundreds of submissions monthly. Poor figure quality signals careless work, even when your data is groundbreaking.\n","content":"Your Matplotlib Plots Are Getting Your Papers Rejected — Here\u0026amp;rsquo;s the Fix You\u0026amp;rsquo;ve spent weeks perfecting your research, written a solid manuscript, and submitted to your target journal. Then the rejection email arrives: \u0026amp;ldquo;Figures do not meet publication standards.\u0026amp;rdquo;\nThe problem isn\u0026amp;rsquo;t your science—it\u0026amp;rsquo;s your pixelated plots, inconsistent fonts, and low-resolution JPEGs. Journal editors see hundreds of submissions monthly. Poor figure quality signals careless work, …","tags":["Matplotlib","Publication figures","rcParams","LaTeX rendering","Vector export","Research visualization","Journal submission","Data visualization"],"section":"blog"},{"title":"Mermaid: Generate Diagrams from Plain Text","permalink":"https://youvenz.github.io/blog/2026-03-04-mermaid-generate-diagrams-from-plain-text/","summary":"Stop Fighting with Diagram Tools — Generate Professional Visuals from Plain Text You need a sequence diagram for your API docs. Opening Lucidchart or draw.io means the next 45 minutes disappear into dragging boxes, nudging arrows three pixels left, and cursing the auto-layout that keeps \u0026ldquo;helping\u0026rdquo; by rearranging everything. Meanwhile, your documentation deadline isn\u0026rsquo;t moving.\nMermaid lets you skip all of that. Write graph TD; A--\u0026gt;B; in plain text, run one command, and get a publication-ready diagram. No mouse, no layout fights, no design decisions.\n","content":"Stop Fighting with Diagram Tools — Generate Professional Visuals from Plain Text You need a sequence diagram for your API docs. Opening Lucidchart or draw.io means the next 45 minutes disappear into dragging boxes, nudging arrows three pixels left, and cursing the auto-layout that keeps \u0026amp;ldquo;helping\u0026amp;rdquo; by rearranging everything. Meanwhile, your documentation deadline isn\u0026amp;rsquo;t moving.\nMermaid lets you skip all of that. Write graph TD; A--\u0026amp;gt;B; in plain text, run one command, and get a …","tags":["Mermaid","diagram generation","text-based diagramming","flowcharts","sequence diagrams","API documentation","technical writing","Inkscape"],"section":"blog"},{"title":"Obsidian Advanced Slides: Markdown Presentations for Researchers","permalink":"https://youvenz.github.io/blog/2026-03-04-obsidian-advanced-slides-markdown-presentations-for-research/","summary":"Create Research Presentations Using Markdown — Inside Obsidian for Academics You\u0026rsquo;re preparing for your advisor meeting in two hours. Your research notes are in Obsidian, your code is in GitHub, and your equations are scattered across three LaTeX files—but now you need slides. You open PowerPoint, spend twenty minutes fighting with equation formatting, realize you can\u0026rsquo;t version control a .pptx file, and wonder why your presentation workflow is stuck in 2005.\n","content":"Create Research Presentations Using Markdown — Inside Obsidian for Academics You\u0026amp;rsquo;re preparing for your advisor meeting in two hours. Your research notes are in Obsidian, your code is in GitHub, and your equations are scattered across three LaTeX files—but now you need slides. You open PowerPoint, spend twenty minutes fighting with equation formatting, realize you can\u0026amp;rsquo;t version control a .pptx file, and wonder why your presentation workflow is stuck in 2005.\nThere\u0026amp;rsquo;s a better …","tags":["Obsidian Advanced Slides","Markdown presentations","Reveal.js","Academic writing tools","Research workflow","LaTeX in presentations","Git version control","Obsidian plugins"],"section":"blog"},{"title":"Overleaf Dark Mode \u0026 Markdown Paste: Hidden Features","permalink":"https://youvenz.github.io/blog/2026-03-04-overleaf-dark-mode-markdown-paste-hidden-features/","summary":"You\u0026rsquo;re staring at a blinding white PDF preview at 11 PM, trying to finish your paper. Your eyes hurt. You\u0026rsquo;ve got structured notes in Obsidian or ChatGPT output in markdown, but you\u0026rsquo;re manually retyping section headers into LaTeX syntax. There\u0026rsquo;s a better way—and it\u0026rsquo;s already built into Overleaf.\nTwo Hidden Features That Actually Matter PDF dark mode inverts your preview to a dark background without touching your exported document. Visual editor markdown paste converts markdown structure—headings, lists, formatting—directly into LaTeX when you paste. Both work on free accounts. No extensions required.\n","content":"You\u0026amp;rsquo;re staring at a blinding white PDF preview at 11 PM, trying to finish your paper. Your eyes hurt. You\u0026amp;rsquo;ve got structured notes in Obsidian or ChatGPT output in markdown, but you\u0026amp;rsquo;re manually retyping section headers into LaTeX syntax. There\u0026amp;rsquo;s a better way—and it\u0026amp;rsquo;s already built into Overleaf.\nTwo Hidden Features That Actually Matter PDF dark mode inverts your preview to a dark background without touching your exported document. Visual editor markdown paste …","tags":["Overleaf","LaTeX","Markdown","Dark Mode","Academic Writing","Research Tools","PDF Preview","Visual Editor"],"section":"blog"},{"title":"TeXstudio AI Macros: GPT-4 Inside LaTeX","permalink":"https://youvenz.github.io/blog/2026-03-04-texstudio-ai-macros-gpt-4-inside-latex/","summary":"Install AI Writing Macros Inside TeXstudio — For Academic Researchers and Technical Writers Series Navigation: This is Part 1 of the LaTeX AI Assistant series. Part 2 (video generation with AI) coming soon.\nYou\u0026rsquo;re deep in writing a research paper when you need to expand a paragraph, explain a complex equation, or generate a methods section from scratch. Instead of context-switching to ChatGPT, copying text back and forth, and reformatting everything — what if you could invoke GPT-4 directly inside your LaTeX editor with a keyboard shortcut?\n","content":"Install AI Writing Macros Inside TeXstudio — For Academic Researchers and Technical Writers Series Navigation: This is Part 1 of the LaTeX AI Assistant series. Part 2 (video generation with AI) coming soon.\nYou\u0026amp;rsquo;re deep in writing a research paper when you need to expand a paragraph, explain a complex equation, or generate a methods section from scratch. Instead of context-switching to ChatGPT, copying text back and forth, and reformatting everything — what if you could invoke GPT-4 …","tags":["TeXstudio","LaTeX","GPT-4","OpenAI API","AI writing assistant","research automation","macro programming","academic writing tools"],"section":"blog"},{"title":"Tool Calling Explained: Turn Your LLM into an AI Agent","permalink":"https://youvenz.github.io/blog/2026-03-04-tool-calling-explained-turn-your-llm-into-an-ai-agent/","summary":"Tool Calling Explained — How to Turn Your LLM into an AI Agent That Actually Does Things Out-of-the-box LLMs can\u0026rsquo;t check your calendar, pull live weather data, or query your database. They\u0026rsquo;re brilliant conversationists trapped in a sensory deprivation chamber, completely isolated from the real world. The result? You get impressive prose about what to do, but zero ability to actually do it.\nTool calling (also called function calling) changes everything. It\u0026rsquo;s the bridge that transforms a chatbot into an agent—an LLM that can invoke external functions and APIs. Yet most explanations overcomplicate it.\n","content":"Tool Calling Explained — How to Turn Your LLM into an AI Agent That Actually Does Things Out-of-the-box LLMs can\u0026amp;rsquo;t check your calendar, pull live weather data, or query your database. They\u0026amp;rsquo;re brilliant conversationists trapped in a sensory deprivation chamber, completely isolated from the real world. The result? You get impressive prose about what to do, but zero ability to actually do it.\nTool calling (also called function calling) changes everything. It\u0026amp;rsquo;s the bridge that …","tags":["Tool Calling","Function Calling","LLM Agents","AI Agents","Large Language Models","API Integration","AI/ML Development","Prompt Engineering"],"section":"blog"},{"title":"What is RAG? Retrieval Augmented Generation Explained","permalink":"https://youvenz.github.io/blog/2026-03-04-what-is-rag-retrieval-augmented-generation-explained/","summary":"RAG Explained — How to Give Your LLM a Memory Without Retraining You\u0026rsquo;ve probably noticed that ChatGPT doesn\u0026rsquo;t know about events from last week, or that your company\u0026rsquo;s fine-tuned model can\u0026rsquo;t answer questions about your internal documentation. Most people assume the solution is retraining the model with new data—an expensive, time-consuming process requiring GPU clusters and ML expertise.\nThere\u0026rsquo;s a better way. LLMs don\u0026rsquo;t actually need to \u0026ldquo;learn\u0026rdquo; new information to use it effectively. They just need access to it at the right moment. That\u0026rsquo;s the insight behind RAG (Retrieval Augmented Generation), and it\u0026rsquo;s why you\u0026rsquo;re seeing it everywhere from customer support bots to research assistants.\n","content":"RAG Explained — How to Give Your LLM a Memory Without Retraining You\u0026amp;rsquo;ve probably noticed that ChatGPT doesn\u0026amp;rsquo;t know about events from last week, or that your company\u0026amp;rsquo;s fine-tuned model can\u0026amp;rsquo;t answer questions about your internal documentation. Most people assume the solution is retraining the model with new data—an expensive, time-consuming process requiring GPU clusters and ML expertise.\nThere\u0026amp;rsquo;s a better way. LLMs don\u0026amp;rsquo;t actually need to \u0026amp;ldquo;learn\u0026amp;rdquo; new …","tags":["RAG","Retrieval Augmented Generation","LLM","Vector Embeddings","LangChain","Semantic Search","AI Practitioners","Vector Databases"],"section":"blog"},{"title":"Write Research Papers in Markdown + Pandoc","permalink":"https://youvenz.github.io/blog/2026-03-04-write-research-papers-in-markdown-pandoc/","summary":"You\u0026rsquo;re staring at a LaTeX error message for the 47th time today. Your paper deadline is tomorrow, but you\u0026rsquo;re debugging \\begin{figure} placement instead of refining your argument. There\u0026rsquo;s a better way: write in clean Markdown, get publication-ready PDFs with equations, cross-references, and IEEE/Springer formatting—all without touching LaTeX syntax until the final export.\nWhat This Workflow Replaces Direct LaTeX editing becomes Markdown + Pandoc conversion. You write in readable .md files with simple syntax for headings, citations, and figures. Pandoc (a universal document converter) transforms your Markdown into professional PDFs or LaTeX source files, using pandoc-crossref for numbered references and citeproc for bibliographies. Output matches journal templates—single-column, two-column IEEE, ACM formats—without manual \\documentclass configuration.\n","content":"You\u0026amp;rsquo;re staring at a LaTeX error message for the 47th time today. Your paper deadline is tomorrow, but you\u0026amp;rsquo;re debugging \\begin{figure} placement instead of refining your argument. There\u0026amp;rsquo;s a better way: write in clean Markdown, get publication-ready PDFs with equations, cross-references, and IEEE/Springer formatting—all without touching LaTeX syntax until the final export.\nWhat This Workflow Replaces Direct LaTeX editing becomes Markdown + Pandoc conversion. You write in readable …","tags":["Pandoc","Markdown","Academic Writing","LaTeX","pandoc-crossref","Research Papers","Citation Management","PDF Generation"],"section":"blog"},{"title":"MedFlowAssist","permalink":"https://youvenz.github.io/projects/medgemma-workflow/","summary":"Overview MedFlowAssist is a multi-agent medical assistant that combines the power of MedGemma, medical speech recognition (MedASR), and LiteLLM to deliver clinical-grade AI assistance across a wide range of healthcare workflows.\nKey Features 13 clinical workflows covering documentation, triage, diagnosis support, and more 12 pre-built scenarios ready to deploy out of the box Voice dictation via MedASR for hands-free clinical note-taking Multimodal input — image, PDF, and CSV upload supported Multi-agent orchestration with LangGraph-style task delegation Clinical Workflows The assistant covers end-to-end workflows including:\n","content":"Overview MedFlowAssist is a multi-agent medical assistant that combines the power of MedGemma, medical speech recognition (MedASR), and LiteLLM to deliver clinical-grade AI assistance across a wide range of healthcare workflows.\nKey Features 13 clinical workflows covering documentation, triage, diagnosis support, and more 12 pre-built scenarios ready to deploy out of the box Voice dictation via MedASR for hands-free clinical note-taking Multimodal input — image, PDF, and CSV upload supported …","tags":["agentic-ai","medgemma","healthcare","llm","python","flask","multi-agent"],"section":"projects"},{"title":"Agentic AI for Health","permalink":"https://youvenz.github.io/research/agentic-ai-for-health/","summary":"Overview Agentic AI refers to systems where a large language model (LLM) acts not just as a question-answering endpoint, but as an autonomous agent — planning, calling tools, retrieving information, and completing multi-step tasks without constant human intervention. In healthcare, this paradigm opens the door to AI systems that can assist clinical workflows, automate research tasks, and synthesise evidence at scale.\nFrameworks I work with LangGraph LangGraph models agent workflows as stateful directed graphs, enabling complex multi-step reasoning chains with loops, conditional branches, and human-in-the-loop checkpoints. This is particularly useful for clinical decision support pipelines that require evidence retrieval → reasoning → recommendation → validation cycles.\n","content":"Overview Agentic AI refers to systems where a large language model (LLM) acts not just as a question-answering endpoint, but as an autonomous agent — planning, calling tools, retrieving information, and completing multi-step tasks without constant human intervention. In healthcare, this paradigm opens the door to AI systems that can assist clinical workflows, automate research tasks, and synthesise evidence at scale.\nFrameworks I work with LangGraph LangGraph models agent workflows as stateful …","tags":["agentic-ai","llm","rag","healthcare"],"section":"research"},{"title":"Benchmark \u0026 Challenges","permalink":"https://youvenz.github.io/research/benchmark-and-challenges/","summary":"Overview Progress in medical AI is only reproducible when the community shares standardised datasets, evaluation protocols, and leaderboards. I contribute to this infrastructure by co-organising international challenges and building open benchmarks that allow fair comparison of methods under identical conditions.\nMARIO Challenge — MICCAI 2024 The MARIO (Monitoring Age-Related Macular Degeneration Intelligence and Outcomes) challenge was a satellite event at MICCAI 2024 in Marrakesh. It provided:\nA multi-modal longitudinal OCT dataset for AMD progression prediction Standardised train/validation/test splits with held-out labels for fair evaluation Two tracks: binary conversion prediction and interval-to-conversion regression Participation from international teams across 3 continents Results were presented at the MICCAI 2024 main conference. The challenge data and evaluation server remain publicly available for continued benchmarking.\n","content":"Overview Progress in medical AI is only reproducible when the community shares standardised datasets, evaluation protocols, and leaderboards. I contribute to this infrastructure by co-organising international challenges and building open benchmarks that allow fair comparison of methods under identical conditions.\nMARIO Challenge — MICCAI 2024 The MARIO (Monitoring Age-Related Macular Degeneration Intelligence and Outcomes) challenge was a satellite event at MICCAI 2024 in Marrakesh. It provided: …","tags":["benchmark","challenge","dataset","retinal"],"section":"research"},{"title":"D2 Ink — D2 Architecture Diagrams for Inkscape","permalink":"https://youvenz.github.io/projects/inkscape-d2/","summary":"Overview D2 Ink integrates the D2 diagram language into Inkscape. D2 produces beautiful, automatically-laid-out architecture diagrams, entity-relationship diagrams, and software design maps from a declarative text syntax.\nWhy D2? D2\u0026rsquo;s automatic layout engine (powered by ELK, Dagre, or TALA) produces publication-quality diagrams without manual positioning. Combined with Inkscape\u0026rsquo;s SVG editing capabilities, D2 Ink is perfect for creating and then polishing technical architecture figures for papers and presentations.\nExample User -\u0026gt; API Gateway -\u0026gt; Auth Service API Gateway -\u0026gt; Business Logic Business Logic -\u0026gt; Database Installation git clone https://github.com/YouvenZ/D2_ink Requires D2 CLI to be on your PATH.\n","content":"Overview D2 Ink integrates the D2 diagram language into Inkscape. D2 produces beautiful, automatically-laid-out architecture diagrams, entity-relationship diagrams, and software design maps from a declarative text syntax.\nWhy D2? D2\u0026amp;rsquo;s automatic layout engine (powered by ELK, Dagre, or TALA) produces publication-quality diagrams without manual positioning. Combined with Inkscape\u0026amp;rsquo;s SVG editing capabilities, D2 Ink is perfect for creating and then polishing technical architecture …","tags":["inkscape","extension","d2","diagrams","architecture","python","open-source"],"section":"projects"},{"title":"Explainable AI for Health","permalink":"https://youvenz.github.io/research/explainable-ai-for-health/","summary":"Overview Deep learning models used in clinical settings must not only be accurate — they must be interpretable. Clinicians need to understand why a model predicts high-risk progression to trust and act on that prediction. Explainable AI (XAI) provides tools to open the black box and reveal which input features drive each decision.\nTechniques I apply Saliency maps Gradient-based saliency methods (GradCAM, Integrated Gradients, SHAP for images) highlight which spatial regions of a retinal image or OCT scan most influenced the model\u0026rsquo;s prediction. A heatmap overlaid on a fundus image showing drusen, haemorrhages, or neovascularisation has direct clinical meaning.\n","content":"Overview Deep learning models used in clinical settings must not only be accurate — they must be interpretable. Clinicians need to understand why a model predicts high-risk progression to trust and act on that prediction. Explainable AI (XAI) provides tools to open the black box and reveal which input features drive each decision.\nTechniques I apply Saliency maps Gradient-based saliency methods (GradCAM, Integrated Gradients, SHAP for images) highlight which spatial regions of a retinal image or …","tags":["xai","interpretability","deep-learning"],"section":"research"},{"title":"ImageGen Ink — AI Image Generator for Inkscape","permalink":"https://youvenz.github.io/projects/inkscape-imagegen/","summary":"Overview ImageGen Ink adds AI image generation to Inkscape\u0026rsquo;s toolset. Describe an image in natural language, choose a backend (Stable Diffusion, DALL·E 3, Flux), and the result is embedded as an SVG \u0026lt;image\u0026gt; element on your canvas — ready to be combined with vector elements.\nBackends supported Stable Diffusion (local via Automatic1111 or ComfyUI API) DALL·E 3 (OpenAI API) Flux (via Replicate or local) Typical workflow Open a new layer in Inkscape for raster assets Run Extensions → ImageGen Ink → Generate Type your prompt (e.g. \u0026ldquo;microscopy image of retinal fundus, professional photo\u0026rdquo;) Choose resolution and backend Image is placed on canvas; scale and position as needed Installation git clone https://github.com/YouvenZ/Imagegen_ink ","content":"Overview ImageGen Ink adds AI image generation to Inkscape\u0026amp;rsquo;s toolset. Describe an image in natural language, choose a backend (Stable Diffusion, DALL·E 3, Flux), and the result is embedded as an SVG \u0026amp;lt;image\u0026amp;gt; element on your canvas — ready to be combined with vector elements.\nBackends supported Stable Diffusion (local via Automatic1111 or ComfyUI API) DALL·E 3 (OpenAI API) Flux (via Replicate or local) Typical workflow Open a new layer in Inkscape for raster assets Run Extensions → …","tags":["inkscape","extension","image-generation","stable-diffusion","dalle","ai","python","open-source"],"section":"projects"},{"title":"Inkscape Extensions Suite","permalink":"https://youvenz.github.io/projects/inkscape-extensions/","summary":"Overview A suite of 8 open-source Inkscape extensions designed for researchers and technical creators who want to stay inside their SVG workflow without switching to external tools.\nEach extension is independently installable and integrates natively into Inkscape\u0026rsquo;s Extensions menu.\nExtensions Diagram Generation SVG Maker — Generate SVG elements from natural language prompts using an LLM, directly inside Inkscape. Mermaid Ink — Render Mermaid.js diagrams (flowcharts, sequence diagrams, Gantt charts) as native SVG inside Inkscape. D2 Ink — Write and render D2 architecture diagrams and entity-relationship diagrams as editable SVG. AI-Assisted Content TextGen Ink — Generate, rewrite, or refine text elements using an LLM — captions, labels, descriptions — without leaving Inkscape. ImageGen Ink — Generate AI images (Stable Diffusion, DALL·E, Flux) from text prompts as embedded SVG elements. Scientific \u0026amp; Data Visualisation Plt Ink — Write Python Matplotlib code and render publication-quality figures as native SVG directly inside Inkscape. Poster Utils Ink — Auto-generate formatted title blocks, author lists, and institution panels for academic conference posters. Academic Reference LoadRefs Ink — Import bibliography files (.bib, .ris, .json, .enw) and place formatted citations as editable SVG text inside Inkscape. Installation Each extension follows the standard Inkscape extension installation process:\n","content":"Overview A suite of 8 open-source Inkscape extensions designed for researchers and technical creators who want to stay inside their SVG workflow without switching to external tools.\nEach extension is independently installable and integrates natively into Inkscape\u0026amp;rsquo;s Extensions menu.\nExtensions Diagram Generation SVG Maker — Generate SVG elements from natural language prompts using an LLM, directly inside Inkscape. Mermaid Ink — Render Mermaid.js diagrams (flowcharts, sequence diagrams, …","tags":["inkscape","extensions","python","svg","ai","productivity","open-source"],"section":"projects"},{"title":"LoadRefs Ink — Bibliography Import for Inkscape","permalink":"https://youvenz.github.io/projects/inkscape-loadrefs/","summary":"Overview LoadRefs Ink solves the citation formatting problem in Inkscape posters and presentations. Import your .bib, .ris, or .json bibliography file, choose a citation style (APA, IEEE, Vancouver, custom), and insert formatted references as native SVG text elements — no copy-paste, no manual formatting.\nSupported input formats Format Extension BibTeX .bib RIS .ris CSL-JSON .json Endnote XML .enw Citation styles APA 7th edition IEEE Vancouver (numerical, for biomedical) Nature Custom CSL template support Workflow Run Extensions → LoadRefs Ink → Insert References Browse to your .bib file Select entries from the parsed list Choose citation style Click Insert — references are placed as a text group at the bottom of your canvas Installation git clone https://github.com/YouvenZ/Loadrefs_Ink pip install pybtex citeproc-py ","content":"Overview LoadRefs Ink solves the citation formatting problem in Inkscape posters and presentations. Import your .bib, .ris, or .json bibliography file, choose a citation style (APA, IEEE, Vancouver, custom), and insert formatted references as native SVG text elements — no copy-paste, no manual formatting.\nSupported input formats Format Extension BibTeX .bib RIS .ris CSL-JSON .json Endnote XML .enw Citation styles APA 7th edition IEEE Vancouver (numerical, for biomedical) Nature Custom CSL …","tags":["inkscape","extension","bibliography","latex","academic","python","open-source"],"section":"projects"},{"title":"Longitudinal Deep Learning","permalink":"https://youvenz.github.io/research/longitudinal-deep-learning/","summary":"Overview Longitudinal learning models the temporal evolution of a patient\u0026rsquo;s condition from sequences of observations collected over months or years. Unlike standard classification or segmentation tasks, longitudinal models must handle irregular time intervals, missing visits, and the inherent continuity of biological processes. My PhD research produced several architectures specifically designed for this setting.\nKey contributions LatiM — Latent Time Models LatiM is a continuous-time latent variable model that represents disease state as a trajectory in a learned latent space. A Neural ODE governs how the latent state evolves between observations, enabling prediction at arbitrary future time points without discretising the timeline.\n","content":"Overview Longitudinal learning models the temporal evolution of a patient\u0026amp;rsquo;s condition from sequences of observations collected over months or years. Unlike standard classification or segmentation tasks, longitudinal models must handle irregular time intervals, missing visits, and the inherent continuity of biological processes. My PhD research produced several architectures specifically designed for this setting.\nKey contributions LatiM — Latent Time Models LatiM is a continuous-time …","tags":["longitudinal","neural-ode","progression"],"section":"research"},{"title":"Medical Image Analysis","permalink":"https://youvenz.github.io/research/medical-image-analysis/","summary":"Overview Medical image analysis is the computational backbone of my research. Before building longitudinal or predictive models, we need robust methods for understanding what is in a single image: detecting lesions, quantifying biomarkers, segmenting anatomical structures. In my work the primary modalities are ophthalmological.\nImaging modalities Colour Fundus Photography Wide-field retinal photographs capture the optic disc, macula, vessels, and peripheral retina. DR grading, vessel segmentation, and optic disc detection are well-established tasks. I use fundus images as the primary modality in several longitudinal and progression prediction pipelines.\n","content":"Overview Medical image analysis is the computational backbone of my research. Before building longitudinal or predictive models, we need robust methods for understanding what is in a single image: detecting lesions, quantifying biomarkers, segmenting anatomical structures. In my work the primary modalities are ophthalmological.\nImaging modalities Colour Fundus Photography Wide-field retinal photographs capture the optic disc, macula, vessels, and peripheral retina. DR grading, vessel …","tags":["medical-imaging","oct","ophthalmology","fundus"],"section":"research"},{"title":"Mermaid Ink — Mermaid Diagrams for Inkscape","permalink":"https://youvenz.github.io/projects/inkscape-mermaid/","summary":"Overview Mermaid Ink brings Mermaid.js diagram syntax directly into Inkscape. Write a flowchart, sequence diagram, Gantt chart, or entity-relationship diagram in Mermaid\u0026rsquo;s plain-text syntax and render it as fully editable SVG — no browser, no external service.\nSupported diagram types Flowcharts (graph TD, graph LR) Sequence diagrams Gantt charts Entity-relationship diagrams Class diagrams State diagrams Installation git clone https://github.com/YouvenZ/mermaid_ink Requires the Mermaid CLI (npm install -g @mermaid-js/mermaid-cli).\n","content":"Overview Mermaid Ink brings Mermaid.js diagram syntax directly into Inkscape. Write a flowchart, sequence diagram, Gantt chart, or entity-relationship diagram in Mermaid\u0026amp;rsquo;s plain-text syntax and render it as fully editable SVG — no browser, no external service.\nSupported diagram types Flowcharts (graph TD, graph LR) Sequence diagrams Gantt charts Entity-relationship diagrams Class diagrams State diagrams Installation git clone https://github.com/YouvenZ/mermaid_ink Requires the Mermaid CLI …","tags":["inkscape","extension","mermaid","diagrams","python","open-source"],"section":"projects"},{"title":"Multi-modal Learning","permalink":"https://youvenz.github.io/research/multi-modal-learning/","summary":"Overview Multi-modal learning addresses one of the fundamental challenges in medical AI: clinical decisions are rarely made from a single data source. A clinician diagnosing diabetic macular oedema consults fundus photographs, OCT B-scans, fluorescein angiography, and the patient\u0026rsquo;s longitudinal record simultaneously. My research develops deep learning architectures that can fuse these heterogeneous modalities into a coherent representation.\nKey research directions Cross-modal feature alignment Standard concatenation of modality-specific features often fails because different modalities live in incompatible representation spaces. I explore contrastive objectives and cross-attention mechanisms that align representations across modalities without requiring paired data at every follow-up visit.\n","content":"Overview Multi-modal learning addresses one of the fundamental challenges in medical AI: clinical decisions are rarely made from a single data source. A clinician diagnosing diabetic macular oedema consults fundus photographs, OCT B-scans, fluorescein angiography, and the patient\u0026amp;rsquo;s longitudinal record simultaneously. My research develops deep learning architectures that can fuse these heterogeneous modalities into a coherent representation.\nKey research directions Cross-modal feature …","tags":["multi-modal","fusion","medical-imaging"],"section":"research"},{"title":"Plt Ink — Matplotlib Figures for Inkscape","permalink":"https://youvenz.github.io/projects/inkscape-plt/","summary":"Overview Plt Ink embeds a Python/Matplotlib code editor inside Inkscape. Write your figure code, hit render, and the resulting SVG is placed on your canvas as a native, fully editable SVG group — not a rasterised image.\nWhy native SVG matters Unlike exporting a PNG from Jupyter and importing it, Plt Ink produces editable SVG: individual lines, bars, and labels are XML elements you can select, recolour, or reposition in Inkscape without going back to Python.\n","content":"Overview Plt Ink embeds a Python/Matplotlib code editor inside Inkscape. Write your figure code, hit render, and the resulting SVG is placed on your canvas as a native, fully editable SVG group — not a rasterised image.\nWhy native SVG matters Unlike exporting a PNG from Jupyter and importing it, Plt Ink produces editable SVG: individual lines, bars, and labels are XML elements you can select, recolour, or reposition in Inkscape without going back to Python.\nFeatures Full Python editor with …","tags":["inkscape","extension","matplotlib","python","data-visualisation","scientific","open-source"],"section":"projects"},{"title":"Poster Utils Ink — Academic Poster Generator for Inkscape","permalink":"https://youvenz.github.io/projects/inkscape-poster-utils/","summary":"Overview Poster Utils Ink automates the most tedious parts of academic poster creation. Fill in your title, authors, affiliations, and abstract — and the extension generates a properly formatted header block, author list with superscript affiliation numbers, and institution row, all placed on the Inkscape canvas at the correct position and typography.\nFeatures Title block at A0/A1/custom poster dimensions Author list with numbering, affiliations, and equal-contribution markers Institution row with logo placeholder slots Abstract section with proper typographic treatment One-click update: change any field and re-render in place Typical workflow For a MICCAI or NeurIPS poster:\n","content":"Overview Poster Utils Ink automates the most tedious parts of academic poster creation. Fill in your title, authors, affiliations, and abstract — and the extension generates a properly formatted header block, author list with superscript affiliation numbers, and institution row, all placed on the Inkscape canvas at the correct position and typography.\nFeatures Title block at A0/A1/custom poster dimensions Author list with numbering, affiliations, and equal-contribution markers Institution row …","tags":["inkscape","extension","poster","academic","python","open-source"],"section":"projects"},{"title":"Predictive Medicine","permalink":"https://youvenz.github.io/research/predictive-medicine/","summary":"Overview Predictive medicine applies machine learning to answer the clinical question: what will happen to this patient? Rather than diagnosing the current state, predictive models estimate the probability and timeline of future events — disease progression, treatment response, or transition to a more severe stage.\nClinical applications Diabetic Retinopathy Progression Diabetic retinopathy is the leading cause of blindness in working-age adults globally. My research builds end-to-end deep learning pipelines that process sequential retinal fundus images and OCT scans to predict whether a patient will progress to proliferative DR or develop diabetic macular oedema within a given follow-up window.\n","content":"Overview Predictive medicine applies machine learning to answer the clinical question: what will happen to this patient? Rather than diagnosing the current state, predictive models estimate the probability and timeline of future events — disease progression, treatment response, or transition to a more severe stage.\nClinical applications Diabetic Retinopathy Progression Diabetic retinopathy is the leading cause of blindness in working-age adults globally. My research builds end-to-end deep …","tags":["diabetic-retinopathy","amd","prediction"],"section":"research"},{"title":"Self-Supervised Learning","permalink":"https://youvenz.github.io/research/self-supervised-learning/","summary":"Overview Supervised deep learning requires large amounts of labelled data. In medical imaging, expert annotations are expensive, time-consuming, and often in short supply. Self-supervised learning (SSL) sidesteps this bottleneck by learning rich representations from unlabelled data through pretext tasks — the labels emerge from the data itself.\nMethods Masked Autoencoders (MAE) Inspired by BERT in NLP, MAE randomly masks a high proportion (75%) of image patches and trains an encoder-decoder to reconstruct the missing regions. The encoder learns spatially rich features without any labels. My L-MAE extends this to temporal sequences of medical images, masking across both space and time.\n","content":"Overview Supervised deep learning requires large amounts of labelled data. In medical imaging, expert annotations are expensive, time-consuming, and often in short supply. Self-supervised learning (SSL) sidesteps this bottleneck by learning rich representations from unlabelled data through pretext tasks — the labels emerge from the data itself.\nMethods Masked Autoencoders (MAE) Inspired by BERT in NLP, MAE randomly masks a high proportion (75%) of image patches and trains an encoder-decoder to …","tags":["self-supervised","masked-autoencoder","representation-learning"],"section":"research"},{"title":"SVG Maker — AI SVG Generator for Inkscape","permalink":"https://youvenz.github.io/projects/inkscape-svg-maker/","summary":"Overview SVG Maker is an Inkscape extension that lets you describe any diagram, icon, or figure in plain English and have an LLM generate the corresponding SVG code, which is then inserted directly into your Inkscape canvas.\nFeatures Natural language → SVG generation powered by GPT-4, Claude, or any OpenAI-compatible API Iterative refinement: describe changes to the selected SVG element to modify it Preview before insertion Works with any Inkscape 1.x installation Installation git clone https://github.com/YouvenZ/svg_maker_ink # Copy to your Inkscape extensions folder # Linux: ~/.config/inkscape/extensions/ # Windows: %APPDATA%\\inkscape\\extensions\\ Requirements Inkscape 1.x Python 3.8+ OpenAI / Anthropic API key (set in extension preferences) ","content":"Overview SVG Maker is an Inkscape extension that lets you describe any diagram, icon, or figure in plain English and have an LLM generate the corresponding SVG code, which is then inserted directly into your Inkscape canvas.\nFeatures Natural language → SVG generation powered by GPT-4, Claude, or any OpenAI-compatible API Iterative refinement: describe changes to the selected SVG element to modify it Preview before insertion Works with any Inkscape 1.x installation Installation git clone …","tags":["inkscape","extension","svg","llm","python","ai","open-source"],"section":"projects"},{"title":"TextGen Ink — AI Text Generator for Inkscape","permalink":"https://youvenz.github.io/projects/inkscape-textgen/","summary":"Overview TextGen Ink is an Inkscape extension that connects any text element on your canvas to a large language model. Select a text frame, choose an operation, and let the LLM generate, rewrite, summarise, or translate your content in-place.\nUse cases Caption generation: select a figure and generate a descriptive caption Label refinement: improve the phrasing of axis labels or legend entries Abstract writing: generate a draft abstract from a set of bullet points Translation: translate all text elements in a scientific poster to another language Supported operations Mode Description Generate Write new text from a prompt Rewrite Paraphrase or improve selected text Summarise Condense long text to a shorter form Translate Translate to a target language Installation git clone https://github.com/YouvenZ/textgen_ink ","content":"Overview TextGen Ink is an Inkscape extension that connects any text element on your canvas to a large language model. Select a text frame, choose an operation, and let the LLM generate, rewrite, summarise, or translate your content in-place.\nUse cases Caption generation: select a figure and generate a descriptive caption Label refinement: improve the phrasing of axis labels or legend entries Abstract writing: generate a draft abstract from a set of bullet points Translation: translate all text …","tags":["inkscape","extension","text","llm","ai","python","open-source"],"section":"projects"},{"title":"Auto Publication List","permalink":"https://youvenz.github.io/projects/auto-publication-list/","summary":"Overview Auto Publication List is a lightweight automation pipeline that keeps your publication list up to date without manual intervention. It fetches your Google Scholar profile export and generates both a BibTeX bibliography file and a LaTeX metrics snippet.\nHow It Works Fetch — Downloads your Google Scholar export (BibTeX-like text) using your Scholar user ID Parse — Cleans and normalises the entries into valid BibTeX format Metrics — Extracts citation counts, h-index, and i10-index and writes them to metrics.tex Export — Outputs publications.bib ready to include in any LaTeX document Usage from update_publications import fetch_scholar_publications # Fetch and export fetch_scholar_publications(user_id=\u0026#34;YOUR_SCHOLAR_ID\u0026#34;, output_dir=\u0026#34;./output\u0026#34;) GitHub Actions Integration Add the workflow to your repository to run on a schedule:\n","content":"Overview Auto Publication List is a lightweight automation pipeline that keeps your publication list up to date without manual intervention. It fetches your Google Scholar profile export and generates both a BibTeX bibliography file and a LaTeX metrics snippet.\nHow It Works Fetch — Downloads your Google Scholar export (BibTeX-like text) using your Scholar user ID Parse — Cleans and normalises the entries into valid BibTeX format Metrics — Extracts citation counts, h-index, and i10-index and …","tags":["python","github-actions","google-scholar","bibtex","latex","automation"],"section":"projects"},{"title":"LatiM: Longitudinal Representation Learning in Continuous-Time Models for Medical Image Analysis","permalink":"https://youvenz.github.io/talks/latim-miccai-2024/","summary":"","content":"","tags":null,"section":"talks"},{"title":"MARIO: Monitoring Age-related Macular Degeneration Progression in OCT — Challenge Results","permalink":"https://youvenz.github.io/talks/mario-amd-2024/","summary":"","content":"","tags":null,"section":"talks"},{"title":"MARIO AMD Progression Challenge","permalink":"https://youvenz.github.io/projects/mario-challenge/","summary":"Overview The MARIO AMD Progression Challenge was held at MICCAI 2024 and focused on the automated assessment of Age-related Macular Degeneration (AMD) progression using deep learning.\nWe are pleased to announce the publication of our comprehensive analysis of the challenge results, and the corresponding dataset is now publicly available for the research community.\nResearch Focus This challenge addressed two key clinical questions:\nProgression prediction: Will a patient\u0026rsquo;s AMD progress over the next 12 months? Visual acuity change: Will the patient\u0026rsquo;s visual acuity improve, stabilise, or worsen? Dataset The MARIO dataset provides:\n","content":"Overview The MARIO AMD Progression Challenge was held at MICCAI 2024 and focused on the automated assessment of Age-related Macular Degeneration (AMD) progression using deep learning.\nWe are pleased to announce the publication of our comprehensive analysis of the challenge results, and the corresponding dataset is now publicly available for the research community.\nResearch Focus This challenge addressed two key clinical questions:\nProgression prediction: Will a patient\u0026amp;rsquo;s AMD progress over …","tags":["challenge","benchmark","amd","medical-imaging","deep-learning","miccai"],"section":"projects"},{"title":"About","permalink":"https://youvenz.github.io/about/","summary":"I am Rachid Youven Zeghlache, a Research Engineer \u0026amp; AI Researcher at IMT Atlantique, France. I hold a Ph.D. from the University of Western Brittany (UBO / LATIM), where my thesis focused on longitudinal deep learning for the prediction of diabetic retinopathy progression.\nMy core research sits at the intersection of medical image analysis, multi-modal learning, and longitudinal modelling — building AI systems that can track disease evolution over time from heterogeneous clinical data such as retinal fundus photographs, OCT volumes, and fluorescein angiography. I have co-organised two MICCAI challenges: MARIO 2024 (AMD progression monitoring in OCT) and DIAMOND 2024 (diabetic macular edema onset prediction), and I serve as reviewer for MICCAI, IEEE TMI, Medical Image Analysis, and Computers in Biology and Medicine.\n","content":"I am Rachid Youven Zeghlache, a Research Engineer \u0026amp;amp; AI Researcher at IMT Atlantique, France. I hold a Ph.D. from the University of Western Brittany (UBO / LATIM), where my thesis focused on longitudinal deep learning for the prediction of diabetic retinopathy progression.\nMy core research sits at the intersection of medical image analysis, multi-modal learning, and longitudinal modelling — building AI systems that can track disease evolution over time from heterogeneous clinical data such as …","tags":null,"section":""},{"title":"Augmented Scholars","permalink":"https://youvenz.github.io/newsletter/","summary":"Augmented Scholars is a newsletter and YouTube channel at the crossroads of AI research and medical science — written by a researcher, for researchers.\nEach issue covers:\n📄 Paper deep-dives — key contributions from MICCAI, NeurIPS, ICML, and top medical AI journals, explained intuitively. 🧪 Method breakdowns — how longitudinal learning, multi-modal fusion, and agentic AI pipelines actually work under the hood. 🛠️ Tools \u0026amp; tutorials — hands-on guides for building AI systems for healthcare (LangGraph, AutoGen, PyTorch, Hugo + GitHub Actions). 💡 Research insights — lessons from a PhD and beyond: experimental design, writing, reproducibility, and navigating the publication process. Subscribe on Substack and watch video essays on YouTube.\n","content":"Augmented Scholars is a newsletter and YouTube channel at the crossroads of AI research and medical science — written by a researcher, for researchers.\nEach issue covers:\n📄 Paper deep-dives — key contributions from MICCAI, NeurIPS, ICML, and top medical AI journals, explained intuitively. 🧪 Method breakdowns — how longitudinal learning, multi-modal fusion, and agentic AI pipelines actually work under the hood. 🛠️ Tools \u0026amp;amp; tutorials — hands-on guides for building AI systems for healthcare …","tags":null,"section":""},{"title":"Contact","permalink":"https://youvenz.github.io/contact/","summary":"","content":"","tags":null,"section":""},{"title":"CV","permalink":"https://youvenz.github.io/cv/","summary":"","content":"","tags":null,"section":""},{"title":"LMT: Longitudinal Mixing Training for Disease Progression Prediction","permalink":"https://youvenz.github.io/talks/lmt-miccai-2023/","summary":"","content":"","tags":null,"section":"talks"},{"title":"Longitudinal Self-Supervised Learning using Neural Ordinary Differential Equation","permalink":"https://youvenz.github.io/talks/ssl-neural-ode-2023/","summary":"","content":"","tags":null,"section":"talks"},{"title":"Detection of Diabetic Retinopathy using Longitudinal Self-Supervised Learning","permalink":"https://youvenz.github.io/talks/dr-ssl-omia-2022/","summary":"","content":"","tags":null,"section":"talks"}]