Course director for the graduate, in-person course taken by all PGY1 residents, and the clinical voice behind this project. Dr. Thang identified the gap — absent students and students who learn better from text than audio — and shaped the assistant around Universal Design for Learning so that every lecture could be offered in multiple means of representation.

The human-in-the-loop review protocol reflects his priority: supplemental material can be generated quickly, but clinical accuracy is checked by an educator before any of it reaches students.

Built and operates the locally-hosted assistant. The design priority was that lecture recordings never leave University of Toronto infrastructure — so the pipeline runs entirely on local engines (whisper.cpp and Ollama) with no external API calls.

Also the author of the sibling RAG Proxy project, presented at TechKnowFile 2026, whose companion site this one is modelled on.