I build AI-native products, and I think like a physicist. Co-Founder and CTO of Lucor AI.
Most of what I have built lives behind security or IP walls. Here are a few you can read:
-
giftaro-public: the code behind Giftaro, a production group-gifting web app (Next.js, React, Stripe Connect, Turso/Drizzle). Built with Claude Code, which is how I ship now. Published as a sanitized snapshot of real production code.
-
ThoughtKeeper: an AI-native thought-capture app (Next.js 16, React 19, local-first SQLite, Anthropic + OpenAI). Hybrid full-text and vector search with an agentic, tool-using assistant.
-
muon-field-averaging: physics analysis from Fermilab's Muon g-2 experiment (recognized by the 2026 Breakthrough Prize in Fundamental Physics). It computes the muon-weighted average magnetic field by convolving the measured beam distribution with the field map. Original methods, full systematics treatment (Python, Jupyter).
-
same-source-prob: a paper deriving a closed-form estimator for "did these two records come from the same source?", validated by simulation with reproducible figure code. The probability theory behind entity resolution and record linkage.
