What I build

Where I am now. I'm an engineer at Composio in SF, working on the tooling that AI agents use to actually get things done. Still early days for me here, so more on this soon.

• Agents
• Infra
• TypeScript
• Python

My own thing. A cross-exchange prediction-markets terminal I built solo, start to finish, in under seven weeks, then spent the next several months growing. At its peak it tracked around $120M in daily volume across Polymarket and Kalshi for 5,000+ people, with an arbitrage engine watching 6,900+ paired markets and firing whale alerts in under a second.

• React
• Python
• Supabase
• WebSockets

My first real job out of school. I rebuilt the biggest client's data infrastructure from scratch, squeezed 7M messy rows down to 1M, and doubled our revenue in the process. Then I automated the whole claims lifecycle and shipped a portal that cut support tickets by around 40%.

• React
• Supabase
• Python

A summer in Boston, my first taste of enterprise scale. I rebuilt a tax data pipeline on PostgreSQL, moving over a million records and rewriting thousands of queries along the way.

• PostgreSQL
• SSIS
• SQL

A weekend that turned into two trading bots sharing one execution engine. One chased whale consensus, one traded crypto binaries, and together they turned $20 into $500 in the first week live. I backtested the whale strategy across 12,775 markets (93.5% win rate, and Monte Carlo said 0% chance of blowing up) and wrapped it in real risk infra: Kelly sizing, circuit breakers, a kill switch at 40% drawdown.

• Python
• WebSockets
• Docker
• GCP

I pointed Claude at 11 seasons of NBA and NFL data, handed it 22 custom SQL tools, and let it hunt for a 1 to 3% edge against the market. The biases it dug up ended up shaping how ProfitLabs' arbitrage engine worked.

• Python
• Claude API
• SQL
• Next.js