The Role
Lead embedded assessments of frontier AI labs’ safety practices. Scope evaluations, audit existing experiments, identify coverage gaps, design safety experiments, lead teams of 3–5 evaluators, and own the scientific content of public reports. Between engagements, conduct practice evaluations, train scientific staff, and improve evaluation tooling with engineers. The role requires in-person work at host labs and includes collaboration with AI developers, governments, and civil society organizations.
Summary Generated by Built In
Salary range: $400,000 - $650,000/year + benefits
Description: Transluce is a fast-moving nonprofit research lab building the public tech stack for AI evaluation and oversight. We partner with frontier AI developers, governments, and civil society organizations to evaluate how AI systems behave, from misaligned agent swarms to child safety and mental health. Our evaluations directly underpin policy, including through collaborations with the EU AI Office and Common Sense Media. We are now building a program of embedded evaluations, in which independent teams assess frontier labs’ safety practices with employee-level access.
About the role: As Scientific Lead for Embedded Assessments, you will scope, design, and lead the assessments carried out by a team of 3-5 evaluators embedded at a frontier AI lab. We expect each engagement to last 1-3 months, during which you’ll work on-site at the host lab to assess the coverage of its internal safety stack. This involves identifying the most important open questions about model safety, auditing the lab’s existing evaluations to find gaps in coverage, and designing experiments to investigate risks and failure modes that haven’t been examined. You’ll own the scientific content of each engagement’s public report. Between engagements, you’ll lead practice evaluations on public models, train other scientific staff, and work with our engineering team to improve our tooling.
Core responsibilities:
- Refine the scientific scope of the assessment based on information gained while embedded.
- Review the output of already-conducted experiments or evals within the host organization with the aim of auditing coverage.
- Design and help execute experiments with the aim of evaluating safety practices of the host organization.
- Lead the scientific input into the final report that will be the culmination of the embedded assessment.
It is crucial for this work to happen in person, since much of the benefit of embedded evals comes from working physically within the host company’s facilities.
Preferred qualifications:
We are looking for candidates with an excellent track record of scientific research, ideally in an area adjacent to AI alignment. Other qualities we value for this role:
- Experience building AI evaluations, such as benchmarks, eval suites, or red-teaming work.
- Experience leading or managing researchers, ideally small, talented teams in a scrappy environment.
- Demonstrated ability to produce high-quality results while prioritizing under uncertainty and tight deadlines.
- Writing ability: both technical writing and the ability to communicate results to less technical team members as well as the general public.
- Ability to quickly incorporate a lot of technical context, in the form of code or other technical artifacts, and begin making progress towards a goal.
We are hiring at all levels of experience and would encourage those enthusiastic about the role who do not meet all of the qualifications to apply. We are located in San Francisco and excited to work together in-person. We are open to sponsoring international visas.
Skills Required
- Excellent track record of scientific research, ideally adjacent to AI alignment
- Experience building AI evaluations, including benchmarks, evaluation suites, or red-teaming
- Experience leading or managing researchers, preferably small teams in a scrappy environment
- Ability to produce high-quality results under uncertainty and tight deadlines
- Strong technical writing and ability to communicate results to technical and general audiences
- Ability to quickly understand technical context, code, and other technical artifacts
- Ability to work in person at host labs and in San Francisco
Am I A Good Fit?
Get Personalized Job Insights.
Our AI-powered fit analysis compares your resume with a job listing so you know if your skills & experience align.
Success! Refresh the page to see how your skills align with this role.
The Company
What We Do
Transluce is an independent research lab that builds open, scalable technology for understanding AI systems and steering them in the public interest. Transluce means to shine light through something to reveal its structure. Today’s complex AI systems are difficult to understand—not even experts can reliably predict their behavior once deployed. Given AI's extraordinary consequences on society, we need scalable and open analyses of the capabilities and risks of AI systems. We are building open source, AI-driven tools to understand and analyze AI systems. We will apply these tools to open-weight models, so the world can vet our analyses and improve their reliability. Once our technology has been vetted, we will work with frontier AI labs and governments to ensure that internal assessments reach the same standards as our publicly vetted procedures. Email: [email protected]
.png)
