Member of technical staff at Anthropic, building RL infrastructure to make models better at safety-relevant capabilities like alignment research and pandemic defense. Previously led AI safety talent programs at BlueDot Impact and advised the UK government on AI policy.