My research interests are language models and agents for reasoning, planning, and tool use; AI safety for reliable, interpretable, and aligned systems; and AI for science to accelerate discovery across scientific domains.