Architected a fully automated content pipeline for the VU Amsterdam Deep Learning course, transforming raw lecture materials into interactive, annotated web handouts with zero manual intervention. The system orchestrates GCP services to extract data from Google Slides and YouTube, uses OpenAI's Whisper for high-fidelity lecture transcription, and leverages LLMs for semantic refinement. I engineered custom parsers — including specialized AppleScript modules for Keynote — to render complex slide decks (PowerPoint/Google Slides/Keynote) into web-ready SVGs and animations. The final static site deploys automatically via GitHub Pages through a CI/CD pipeline. The platform serves as the primary learning resource for hundreds of students each year, as can be seen here. Due to the sensitivity of the information the repository is private.