Senior Researcher, CRUX
Build open-world evaluations of frontier AI agents and shape the research agenda for CRUX
CRUX (Collaborative Research for Updating AI eXpectations) conducts open-world evaluations: long-horizon tasks in real-world environments where success cannot be neatly specified or automatically graded. Recent and ongoing evaluations include autonomous iOS app development and publication, and AI R&D automation.
We are hiring a Senior Researcher to co-lead these evaluations. You will shape which questions we investigate, design the studies, and be responsible for delivering published results on a regular cadence. There will be flexibility, especially over time, to work across our broader portfolio of evaluation projects. The role carries substantial independence in setting direction, and you will work closely with the software engineers on our team who build and operate the agent infrastructure that supports each evaluation.
Responsibilities
- Design new open-world evaluations in domains such as AI R&D, software engineering, novel game playing, research tasks across disciplines, and internal governance tasks within enterprises.
- Plan and oversee evaluation runs in coordination with software engineers building agent scaffolds, virtual machines, and supporting infrastructure.
- Analyze agent logs and trajectories to characterize how agents succeed and fail, combining automated tooling with careful manual review.
- Lead report writing and public write-ups on a publication cycle of roughly one to two months.
- Coordinate with external collaborators across academia, government, industry, and civil society throughout study design, execution, and review.
Who we are looking for
We weigh demonstrated work more heavily than credentials, and a PhD is not required. Strong candidates will have:
- A track record of independently driving ambitious technical or empirical research projects from open-ended goals to finished, public results.
- Strong engineering skills, including proficiency in Python, familiarity with cloud infrastructure, and hands-on experience with LLM agents and scaffolds.
- Sound research judgment: the ability to assess ML papers or agent research outputs at the level of a strong conference reviewer.
- A record of delivering research on short timelines while making reasonable tradeoffs between speed and polish.
- Clear, efficient writing for both expert and public audiences.
Compensation and logistics
- Salary: $150,000-$220,000 per year, depending on experience, or market-adjusted salary worldwide.
- Location: fully remote. We can hire from nearly anywhere; international staff are employed through an employer of record.
- Benefits: competitive benefits, including high-quality medical, dental, and vision coverage for you and your family, and up to four months of paid parental leave.
How to apply
Send a CV, a brief statement of interest, and links to relevant work that you substantively led (papers, code, or public writing) using the application form. Applications are reviewed on a rolling basis.
Our process consists of an introductory call, a technical and research interview, and a paid work trial on a CRUX task.
We are an equal opportunity employer and encourage applications from candidates of all backgrounds.