In progress · 2026
LLM validation agent for AI-generated lessons
I own the design of an LLM validation agent that checks AI-generated lessons. So far I have mapped and audited the existing AI generation pipeline and written the design plan for the agent. Next: a shadow-mode prototype, measured against a labeled test set.
LLM-as-judge · Evaluation design · Pipeline audit
Work in progress at Cocoa Classroom. The code is private.