I'm a PhD student at EPFL in my final year, advised by Prof. Maryam Kamgarpour. I am passionate about theory and algorithms for aligning AI models with human objectives. I focus on imitation learning, inverse RL, and preference-based RL, as well as their application to robotics.
🔗 Website • Google Scholar • LinkedIn
Open source projects
- isaac_rlhf: Implements RLHF with randomized exploration for Nvidia IsaacLab tasks.
- isaac_irl: Implements inverse RL methods such as GAIL, Wasserstein Adversarial Imitation Learning, and generic IRL methods for Nvidia IsaacLab tasks.
- tabular_gym: Tabular RL library implementing various algorithms, such as (soft) value/policy iteration, linear programming, natural policy gradient, as well as min-max methods for CMDPs and IRL.
- transfer_irl: Implements code to evaluate transferability of rewards in inverse reinforcement learning.
- cirl: Implements code for constrained inverse reinforcement learning.
- stable-ndde: Implements code for stable neural delay differential equations.