My research interests center around using RL to learn efficient representations for combinatorial reasoning. Currently, I’m investigating any-order autoregression algorithms with self-correcting behavior as well as self-supervised skill learning for agentic workflows. Previously, I completed my Master’s thesis on designing spatial birth-death point processes to model decentralized morphogenesis. I received my M.S.E. from Princeton University and my A.B. from Harvard University, both in computer science.