Description
FAR.AI is hiring a Research Scientist to develop, evaluate, and deploy white-box methods that use model internals to improve AI safety, with a focus on agentic coding, long-context models, realistic threat models, and practical monitors and interventions. The role involves publishing research, engaging with the AI alignment community, and proposing new research directions. Candidates should have hands-on experience with white-box methods, a track record in AI safety or related applied research, and a PhD or several years of research experience. The position is full-time and in-person in Berkeley, California, with visa sponsorship for in-person employees.
