ML Engineer — Reinforcement Learning
Oct 2024 — Present
- Develop and evaluate RL locomotion controllers for quadrupeds deployed in industrial inspection.
- Run end-to-end experiments across GPU simulation, policy training, robustness evaluation, and sim-to-real iteration.
- Design curricula and scenario suites targeting rare locomotion failures.
- Led and secured a EuroHPC grant enabling up to 50,000 H100 GPU-hours for large-scale RL robustness and scaling experiments.