Srijan Sood JP Morgan’s AI Research Group Founding Member
Cornell Financial Engineering Manhattan 2026 Ai & Future of Finance Conference : September 11 2026
Srijan Sood is an AI Research Lead and founding member of J.P. Morgan AI Research, where he established and leads the research program on Foundation Models for Markets & Trading – from pre-training large-scale transformers on billions of raw trade messages, through post-training and RL, to world models of markets.
He architected TradeFM – an autoregressive transformer paired with a deterministic market simulator – which generates high-fidelity synthetic order flow, reproduces key stylized facts of real markets, and transfers zero-shot across temporal and geographic regime shifts. Applications include forecasting, optimal execution, and agent training, through task-specific harnesses.
Srijan’s research also spans deep reinforcement learning for portfolio optimization, knowledge graphs and LLM agents for investment analysis, across 15 publications and 6 U.S. patents, alongside institutional-client adoption and advisory engagements with hedge funds and asset managers.
He studied Computer Science at Georgia Tech (M.Sc., B.Sc.), where he worked on reinforcement learning – language-model alignment with Mark Riedl (pre-RLHF), equities portfolios with Tucker Balch, and cross-domain reward functions with Charles Isbell.
Srijan is now focused on world models of markets – foundation models a trading agent can plan inside.
