$100K multi-asset book, systematic and discretionary, running at a 2.1 Sharpe. Rebuilt the firm's entire research and deployment pipeline.
Tanush Shah
Electrical Engineering, University of Waterloo
Trading a multi-asset book at StageRail Capital and building agentic AI at Petra Lab X. On the side, a C++ matching engine and a prediction-market arbitrage scanner. Before Waterloo, eight years of competitive robotics, ending in a VEX World Championship final.
Work
Architected the company's agentic AI platform: a knowledge graph unifying its core systems, with LLM agents that can trace any transaction across 100k+ live records.
Built a robotic picking system that replaced manual work with a cobot, and shipped it to production.
Projects
A real-time scanner across 10+ sportsbooks, Kalshi and Polymarket that strips out the vig and flags arbitrage.
A C++ matching engine that handles 100k+ orders/sec, with a market maker on top that cut max drawdown 35% in volatile simulations.
Earlier
Led the team to the VEX World Championship finals, top 2 of 20,000 teams worldwide, winning a Worlds division and two New Jersey championships along the way.
Designed, built and programmed a modular large-format gantry 3D printer, end to end.
The only AI team picked for the VEX Worlds showcase. Made the robot's onboard AI 5x faster on a Jetson Nano.