james@stanford ~ % 
     __                     _______
 __ / /__ ___ _  ___ ___   / ___/ /  ___ ___  ___ _
/ // / _ `/  ' \/ -_|_-<  / /__/ _ \/ -_) _ \/ _ `/
\___/\_,_/_/_/_/\__/___/  \___/_//_/\__/_//_/\_, /
                                            /___/

I am a Master's student in Computer Science at Stanford University. Previously, I obtained Bachelor's degrees in Computer Science and Business Administration from UC Berkeley, where I was a Regents' and Chancellor's Scholar.

I am interested in data curation and pipelining for reinforcement learning approaches to reasoning problems across language, vision, and physical domains.

Email Google Scholar GitHub LinkedIn
cat ~/research.md

Research

Follower Agnostic Methods for Stackelberg Games
Chinmay Maheshwari, James Cheng, S Shankar Sastry, Lillian Ratliff, Eric Mazumdar
CDC, 2024
pdf arXiv

Methods for leader-follower (Stackelberg) games with follower-agnostic guarantees.

ls ~/projects/

Projects

Constrained Text Generation Benchmark

How well can LLMs follow structural rules like "write exactly 20 words" or "end each sentence with a rhyme"?

pdf Code
MuSR: Murder Mystery Reasoning Benchmark

Can LLMs update their beliefs and reason counterfactually when solving murder mysteries?

LLM-as-Planner vs RL on Procgen

What happens when you let an LLM plan the strategy and RL execute the actions?