About METR
What is METR?
METR (pronounced "meter") is "a research nonprofit that scientifically measures whether and when AI systems might threaten catastrophic harm to society." The name stands for Model Evaluation and Threat Research. Its stated mission is "to develop scientific methods to assess catastrophic risks stemming from AI systems' autonomous capabilities and enable good decision-making about their development."
Most of its work is evaluations of how far an AI system can "autonomously carry out substantial tasks". Its published research includes a finding that the length of tasks AI agents can complete has doubled about every 7 months for 6 years, an evaluation of GPT-5.1-Codex-Max, and a randomised controlled trial showing that experienced open-source developers overestimate how much AI tools help their productivity. METR also contributed to evaluation standards (STREAM and AEF-1) and prototyped the Responsible Scaling Policies approach, which it says has been adopted by nine AI developers.
Where will I work?
Mostly in person, in downtown Berkeley, California. Every open role on the careers page is marked "On-site", with one policy role in Washington, D.C. For technical roles METR says it "fairly strongly" prefers someone who can be in person, "but WFH some days of the week is fine, and remote is not a complete dealbreaker"; for operations roles it expects candidates "who can relocate and work in-person", and the Head of People posting says "Our operations team members are in our office in Berkeley ~4-5 days/week." METR says it has no corporate backer: it is funded by donations, with a small part of its income from a technical assistance contract with the European AI Office, and "has not accepted funding from AI companies."
What is the METR team like?
The about page lists the founder and CEO (Beth Barnes), a president, a chief scientist and a CTO, followed by technical, operations and policy staff and a set of advisors. Postings describe "a tight-knit, collaborative research culture rooted in truth-seeking and integrity" and "a low-ego, drama-free environment focused on what matters." METR has previously partnered with OpenAI, Anthropic, Google DeepMind, Meta and Amazon to pilot frontier risk assessments, belongs to the NIST AI Safety Institute Consortium, and partners with the AI Security Institute.
Hiring usually combines a code screen, take-home work tests, technical and behavioral interviews and an in-person work trial of about two days; METR says it reimburses travel and lodging and pays for work tests and trials.
Work-Life Balance
The careers page lists "Unlimited PTO" among its benefits, and METR's general posting adds a 21-week parental leave for new parents. Other budgets support life outside the office: a wellness benefit of $7,500 a year, a mental health benefit for therapy and medication, a commuter stipend and rideshare budget, and catered lunch and dinner with an in-office gym and shower. Open-role postings show published salary ranges, from $153K-$215K for an IT Systems Engineer to $492K-$578K for the Chief Information Security Officer.
Perks and Benefits
- Salaries "competitive with top AI labs"
- Medical, dental and vision insurance
- Fertility benefits through Carrot
- Unlimited PTO and a 21-week parental leave for new parents
- Mental health benefit for therapy, medication and other expenses; wellness benefits of $7,500 a year
- Professional development budget of $5,250 a year
- Rideshare benefits of $15k a year
- Management coaching ($2,500+ every 6 months)
- Commuter benefit: monthly transit or parking stipend and an annual Uber budget
- Catered lunch and dinner daily, with an in-office gym and shower
- Relocation support: a stipend for moving to the Bay Area
- Home office and workstation equipment benefit