PayerBench
An environment for evaluating and training agents on insurance claim adjudication. I’m exploring how models gather evidence, predict payer decisions, and learn from feedback.
Research in progress
FOUNDER & ENGINEER · NEW YORK
I build AI agents, evaluation systems, and RL environments.
I’m the co-founder and CTO of Arrow, with 10+ years building healthcare and fintech platforms. I work on production AI agents for medical billing, and the environments and graders that help them improve.
My current focus is reinforcement learning and agent evaluation: turning failures in real workflows into better training environments, feedback, and regression tests.
An environment for evaluating and training agents on insurance claim adjudication. I’m exploring how models gather evidence, predict payer decisions, and learn from feedback.
Research in progress
Built production agents for denials, appeals, payer calls, and EHR workflows, alongside a reset, replay, and grading framework for evaluating and improving them.
Built Walnut, the healthcare lending platform that became Arrow, scaling to three core products and $15M in annual revenue.
Previously, senior software engineer at Charles Schwab and Motif Investing. B.S. in Electrical & Computer Engineering from The Ohio State University.
If you’re working on RL environments, agent evaluation, or healthcare AI, I’d like to connect.