Researcher, Evaluations

• Evaluate frontier AI models on real-world tasks by curating and refining benchmark suites.
• Create and curate an evaluation suite with realistic tasks that challenge practical AI capabilities.
• Design grading rubrics and assess AI performance quantitatively and qualitatively on complex work.
• Communicate research findings through reports, blog posts, and visualizations to inform stakeholders.

This role helps ensure frontier AI models are rigorously evaluated on real-world tasks, contributing to safer and more reliable AI development.

• Experience in AI evaluation, benchmarking, or related research.
• Strong data analysis skills and ability to automate workflows.
• Excellent communication skills for reporting and visualization.

Compensation and benefits not specified in the posting.

Epoch AI is a research organization focused on analyzing and forecasting the trajectory of AI development, particularly in the context of AI safety and governance.