Things I've built — mostly at the intersection of AI, health, and sports.
An n-of-1 wearable health analysis system. It turns long-term WHOOP, Ōura, and Garmin data into statistically valid personal studies — using changepoint detection, multiple-testing-corrected correlation analysis, and a RAG assistant grounded in peer-reviewed health literature — so people can find what actually moves their own physiology, not the average person's. Won Conway's Best AI for Decision Support Award at Carnegie Mellon.
A data-driven college football ranking model — an alternative to journalist-voted polls. It flagged mismatches the AP missed, ranking teams on performance rather than reputation.
A distributed LLM inference network — pooling compute across machines to run large models.
A web app for high school football coaches to design and plan plays, complete with built-in defensive coverages — making advanced game planning accessible to programs with smaller budgets.