What we do
KelAI is an AI research lab building agentic alpha researchAutonomous agents that conduct rigorous quantitative research end to end and produce systematic trading signals., evalsEvaluations quantifying the reliability of AI models on quantitative equity research, graded by experts against institutional criteria. and RL environmentsResearch environments with verifiable rewards that penalize poor research behavior, from methodological flaws to reasoning and calculation errors. to push the frontier of AI in quantitative finance.
Where our founding team comes from
Our research thesis
Building trust in financial AI.
AI has acquired financial knowledge. It reads filings, explains valuation frameworks, and summarizes earnings calls with fluency. Yet finance is, at its core, an industry of forecasting, where every allocation is a view on the future, and forecasting depends on the ability to model how markets behave and evolve.
This is the next frontier for financial AI, and its obstacle is trust rather than capability. Today, portfolio managers still question whether they can rely on the research their agents produce, even as it is clear that AI will one day understand markets more deeply than any human. That tension, between the trust AI holds today and the capability it is heading towards, is what led me to start KelAI.
KelAI brings together specialized research agents, rigorous evaluations of their work, and reinforcement-learning environments that sharpen their research discipline. Together they form a cycle of research, evaluation and improvement, all in service of one mission: to build AI that models markets reliably.
Jeremie Cohen, Founder & CEO
Careers
Join our team in New York
We are a small research team working from New York. We are hiring AI and quantitative researchers.
See open roles