Projects

Four things, not everything I have built.

A live benchmark for AI fluency: how well people actually work with these tools.

Live — fluencybench.com

A tool for finding the gaps in an AI scoring rubric, by showing a team where its own people quietly disagree about what counts as good.

Live — evallab-eosin.vercel.app

Public AI learning system with code, notes, experiments, and proof artifacts.

Live — on this site

Step through how a system learns to reach a goal, one decision at a time. The idea underneath reinforcement learning, without the notation.

Live — policy-iteration.vercel.app