Research Scientist, Applied White-Box Methods
FAR AI · Berkeley Office · Hybrid (US)
CHECKED 19H AGO
Research institutehybrid$150k–$250k
At a glance
Highlights
- Develop white-box methods leveraging model internals to improve AI safety.
- Stress-test interpretability methods on long-context agentic coding tasks.
- Build Goodhart-resistant evaluations to guide automated research iteration.
What they offer
- 94% of health insurance premium paid, starting within 1 month
- 401(k) plan with up to 2% match
- 25 days PTO per year plus up to 10 days paid sick leave
- Paid bereavement, family, medical and pregnancy disability leave
- Work computer and WFH stipend for eligible employees
- Catered lunches and dinners on workdays in Berkeley office
Ideal candidate
A mission-driven researcher with several years of experience in ML or a quantitative science who is excited to bridge early-stage interpretability research to real-world AI safety deployment.
Requirements
- Several years of research experience in computer science, machine learning, physics, statistics, or a related field
- Previous experience in applied ML for other fields: e.g. biology, chemistry, materials science, robotics, etc.
- remote policy
- In-person preferred; remote considered for exceptional candidates
- visa sponsorship
- Visa sponsorship available for in-person employees
- hiring process
- Interviews with members of technical staff, followed by a paid work trial of up to one week
Lab profile
About FAR AI
Non-profit AI research institute ensuring advanced AI is safe and beneficial for everyone.
Known for
- AI safety research
- AI risks and solutions research
- Red teaming
Key facts
- Non-profit research institute