Researcher, Benchmark Reviews

Epoch AI is seeking a Researcher to analyze and review AI benchmarks, producing public-facing reports and evaluations. The role involves critical analysis of AI evaluation methodologies and data, with a focus on transparency and rigor.

Location: Remote (many countries)

Salary: $100,000–$200,000 USD annually, depending on experience and location

Responsibilities

• Review benchmarks: Assess new benchmarks bi-weekly, evaluate methodologies, and interpret AI capabilities.
• Publish and maintain reports: Write and update public reports on benchmarks and model progress.
• Dig into data: Analyze individual tasks within benchmarks, using coding agents with critical oversight.

Requirements

• Excellent written communication skills, producing clear, publication-quality drafts.
• Critical thinking about research methodologies and evaluation standards.
• Familiarity with existing AI benchmarks, their strengths and weaknesses.
• Comfort working with dense data and breaking down benchmarks into components.

Benefits

• Annual salary between $100,000 and $200,000 USD, depending on location, seniority, and experience.
• Fully remote environment with flexible work hours.
• Comprehensive health insurance, life insurance, pension plan (if applicable).
• Generous paid time off, including 30 days PTO, unlimited sick/personal leave, 4 months parental leave.
• Expense policy for equipment and learning tools, including AI tools.
• Paid trips for conferences and retreats.
• Access to well-equipped offices in Berkeley, CA, with meals, gym, and more.

Additional Information

• Applicants from all time zones are welcome, with preference for overlap with US and UK hours.
• Applications should be in English with professional proficiency.
• Epoch AI values diversity and inclusivity, encouraging applicants regardless of background.
• Contact via email for questions or accessibility requests, but applications must be submitted through the provided link.