2026—
Tarski Research — Automated reasoning at YC
What I'm working on now. More on this soon.
Asher Feldman
I'm a Turing Scholar at UT Austin studying computer science and mathematics. I'm currently Chief Scientist and Co-founder of Tarski Research. Before that I worked on automated theorem proving in the UToPiA Lab, where I worked on making LLM-based provers better at Lean 4 by using RL to learn when to retrieve custom tactics.
Research & Work
2026
UToPiA Lab, UT Austin — Undergraduate Researcher
Worked with Prof. Isil Dillig, Maxine Xin, and Prof. Elias Stengel-Eskin: made a dynamic retrieval system that injects relevant custom proof tactics into LLM-based theorem provers at inference time.
2025
Lean 4 Directed Reading Program — Member
Functional programming and formal verification in Lean 4. Proved the termination of quicksort by hand, which is what got me interested in formal reasoning.
Education
2025—29
University of Texas at Austin — CS (Turing) + Mathematics