About me!

Hi! I am a final year PhD student at the University of Texas at Austin, working in AI for mathematics. I am advised by Dr. Swarat Chaudhuri in the Trustworthy & Intelligent Systems Lab. I am also a Student Researcher at Google DeepMind, as part of the Science and Strategic Initiatives Unit. I am always happy to meet new people, please reach out to me if you’d like to chat!

Research

My researcher at UT Austin has been focused on AI techniques for formal mathematics. I organized a team of university students to produce a new formal competition maths benchmark for evaluating the next generation of mathematical reasoning automation. PutnamBench was accepted at NeurIPS 2024 and recognized at the ICML 2024 AI for Math Workshop via the Best Paper Award. Since publication, it has become the leading standard for formal maths evaluation. I also led FERMAT, a project aimed to better understand: how did human mathematics arise? how can we learn a policy that can reproduce human-made (interesting) mathematics, tabula rasa? The FERMAT (Learning Interestingness in Automated Thoery Formation) work was accepted at NeurIPS 2025 as a Spotlight paper.

I’ve also contributed to COPRA, one of the earliest agentic approaches for formal theorem-proving, CLEVER, a benchmark for software verification in Lean, and ProofWala, infrastructure supporting AI for theorem-proving research and experiments regarding multi-language formal data training.

While in undergrad, I conducted research with the Rutgers Automated Reasoning Lab, advised by He Zhu, where I worked on using differentiable programming for invariant synthesis for program verification. Before that, I was involved in research in theoretical maths. I participated in an NSF REU at the College of William & Mary under the advisement of Charles R. Johnson, and at San Diego State University, under the advisement of Chris O’Neill.