Full Professor, University of California, San Diego
1 paper at NeurIPS 2025
We adopt reinforcement learning to train LLMs to generate quality code with rewards derived from program analysis.
