Bluesky

We're trying to develop the underlying methodologies to track AI risk from loss of control. This includes creating ways to measure what AI systems can do autonomously, studying AI ability to evade monitoring, and publishing comprehensive risk assessments.

We've also contributed to standards for evaluation methodology (including STREAM and AEF-1) and helped propose frontier safety policies, a type of governance approach now adopted by most major AI developers.

We invest in exceptional people

Salaries competitive with top AI labs

Medical, dental, and vision insurance

Fertility benefits through Carrot

Unlimited PTO

Wellness benefits ($7,500/year)

Professional development ($5,250/year)

Rideshare benefits ($15k/year)

Management coaching ($2,500+/6mo)

Catered lunch and dinner

What to expect when interviewing

METR's hiring process usually involves a combination or subset of the following:

Code screen
Take-home work test(s)
Technical interview/assessment(s)
Behavioral interview(s)
In-person work trial (typically ~2 days)

We reimburse all travel/lodging expenses, and pay for work tests and work trials (subject to visa and other restrictions).

A note on recruitment fraud. We are aware of people falsely claiming to recruit on METR's behalf. All legitimate communication from METR comes from an @metr.org email address. We never ask candidates to pay for assessments, resume services, training, equipment, or anything else. If you receive an offer, job description, or interview request you're unsure about, please ask hiring@metr.org for confirmation.