Embedded AI Risk Researcher

METR; Berkeley (CA); Hybrid; USD 402,000 - 688,000; Full time

Benefits offered by this job

Catered lunch and dinner daily

Relocation stipend

Unlimited PTO

21-week parental leave

Transit/parking stipend

Professional development

Mental health support

Wellness benefits

Home office equipment

Job summary

METR embeds researchers and engineers inside frontier AI labs for extended periods to investigate incidents, stress-test internal monitoring, and assess loss-of-control risks in deployment. The role evolves as the risk assessment methodology matures, with opportunities to build tooling, scale operations, and contribute to independent information about AI risks.

Ideal candidates bring expertise in prompt engineering, security, or large codebases, plus strong communication and the ability to work

Qualifications

• Two or more of the following skills are required: prompting LLMs, security experience, ability to navigate large codebases, strong communication, threat modeling expertise.

Responsibilities

• Embed in frontier AI labs for weeks at a time with 1–4 METR staff.
• Develop methodology, talk to researchers, build tooling, write up results.

Skills

prompting LLMs

security experience

codebase familiarity

loss-of-control threat modeling

communication & stakeholder mgmt

trustworthy reputation

Job description

METR embeds researchers and engineers inside frontier AI labs for extended periods to investigate incidents, stress-test internal monitoring, and assess loss-of-control risks in deployment. The role evolves as the risk assessment methodology matures, with opportunities to build tooling, scale operations, and contribute to independent information about AI risks.

Ideal candidates bring expertise in prompt engineering, security, or large codebases, plus strong communication and the ability to work