I build agentic AI systems across onchain markets, private capital, and personal AI. I studied economics and media studies at the University of Virginia and am currently pursuing the Master in Management at Kellogg School of Management at Northwestern University. My prior work spans private equity at SP Capital (Hong Kong and Shanghai), venture capital at Jump Investors (Los Angeles), DAO sourcing at Party Ventures (New York), and energy economics research at the University of International Business and Economics (Beijing).
I also shoot 35mm film. The photography lives at fnu.cargo.site.
OPick is a live opinion market on Base mainnet. It auto-classifies 1.2M+ X posts a day through the Anthropic API and settles attention-index markets onchain, with no manual intervention.
Live on Base mainnet · ~60 days continuous · 6 settlement cycles
Spandrel is an agentic deal-intelligence system for a US private-capital firm in the transferable tax credit market. It learns the firm's deal criteria from their own portfolio and scores new deals nightly with rationales the team can override.
~7 actionable deals surfaced per week · running daily in production
Crude Signal tests whether WSJ text carried crude oil signal that futures markets had not yet priced. An LDA topic model on 200K+ articles, built for the UIBE Beijing macro team.
Sharpe 0.07 to 0.45 · forecast error down 18% (out-of-sample)
An agentic trading bot that reads 18 RSS feeds in Chinese and English, runs causal-chain analysis through the Anthropic API, and paper-trades Kalshi event contracts on US economic indicators.
CPI move flagged ahead of the print · 8 of 12 signals correct (paper-traded, 4 months)
Larry is a personal AI assistant I built from scratch and use every day, running inside WhatsApp.
Built from scratch · used every day
Miletus is a protocol design for critical-mineral perpetual swaps, a market with no clean reference price.
Design only, not deployed · 3-source layered oracle