I build tools for robot learning and the software around it. I'm a robotics researcher at USC's Dynamic Robotics and Control Laboratory, studying mechanical engineering and pursuing an M.S. in Artificial Intelligence and Machine Learning.
My work spans reinforcement learning, MuJoCo simulation, motion retargeting, and the tools needed to inspect experiments and evaluate what a policy actually learned.
| Project | What to explore |
|---|---|
| RL-Sculptor | A Python/FastAPI/React platform for generating and revising RL rewards from simulation rollouts, with research retrieval and checks on generated metrics. Demo. |
| Humanoid Harness | Research on revising reference guidance and task rewards for humanoid post-training. The public repository documents its development proxy, experiments, and integration limits. |
| PIT Backtest | A point-in-time equity research framework with leakage checks, explicit cost assumptions, and reproducible figures. |
| Pendulum Drag Extraction | A team experiment estimating drag from pendulum decay, with MATLAB analysis, apparatus details, figures, and the report. |
Tools I work with: Python, C++, TypeScript, PyTorch, MuJoCo/MJX, JAX, React, FastAPI, MATLAB, and CAD.
I'm interested in robot learning, simulation, and practical tools that make experiments easier to reproduce and understand.