HISTORY
& AI

I build LLM systems over large archives of historical and government documents — semantic search, knowledge graphs, and evaluation — and apply them to intelligence-analysis problems.

I'm currently technical co-lead on MIDLLMAT, a project under IARPA's BENGAL program with Columbia's History Lab, measuring mosaic-inference risk: whether an LLM can reconstruct a higher-classified conclusion from only lower-classified documents. That work builds on my research at the History Lab, where I created HistoryLab AI — conversational search over five million declassified FOIA documents.

I believe technology should serve human understanding, not replace it. My senior thesis on WWII strategic bombing — a data-complete reanalysis of every U.S. raid in the European theater — won Columbia's Alan J. Willen Memorial Prize for challenging 80 years of historical narrative with data-driven methods.

I write on Substack about technology, culture, and ideas. My main publication, Political Atheism, applies René Girard's mimetic theory to literature, culture, and contemporary society through close readings, philosophical analysis, and computational modeling.