My research focuses on reinforcement learning, large language models, agentic AI, world models, and AI safety and robustness. I develop methods for reliable sequential decision-making in complex and uncertain environments, with particular interests in efficient inference, adaptive computation, robust learning, long-horizon agent behavior, and multi-agent systems. I am interested in translating these methods into practical AI systems that operate reliably under computational, data, and safety constraints, and in collaborating with industry on both foundational research and applied R&D.