I recently graduated from Carnegie Mellon University where I studied computer science and completed a senior thesis on learning safety constraints from expert demonstrations.
My research interests lie in robust and safe AI, particularly in challenging settings involving interaction or adversaries. Some directions I'm excited about include better specifications of preferences, interpreting large-scale models, and designing and scaling reinforcement learning algorithms. I'm also broadly interested in topics adjacent to multi-agent systems, physical simulation, and designing and optimizing low-level systems.
Notable projects:
multi-task inverse constraint learning
robust fine-tuning of large-scale pretrained models
unsupervised, interpretable image editing
parallel fluid simulation in CUDA
Previously, I've done internships at Jump Trading, Meta AI, and Meta. At Carnegie Mellon, I've been a teaching assistant for 15-251, an introductory course on theoretical computer science.
In my free time, I enjoy playing chess, poker, and going bouldering.