Gold medalist at the International Biology Olympiad
Early work on Energy-Based Models (EBMs) in 2020 contributed to the development of diffusion models
Published influential work across generative modeling, reasoning, perception, and interactive agents
Organizing workshops on 'Physically Grounded World Models' at ICML 2025 and 'World Models in Robotics' at CoRL 2025
PhD thesis focused on compositional world models for embodied intelligence
Background
Assistant Professor at Harvard University's Kempner Institute and Department of Computer Science
Research focuses on generative models, decision making, robot learning, embodied agents, and their applications in scientific domains
Aims to develop intelligent embodied agents that operate in the physical world
Builds compositional generative models (e.g., Energy-Based Models, EBMs) to enable systematic planning and iterative reasoning
Interested in decentralized generative architectures for decision-making, integrating multimodal models (e.g., 3D perception, memory, auditory understanding) that cooperate jointly
Emphasizes the importance of test-time search for effective multimodal and decision-making agents
Explores general improvements to generative modeling and applications in computational biology and inverse design