Researcher, Evaluations and Benchmarks
Core
Design and ship AI safety benchmarks every 2-3 weeks to measure frontier risks for top AI labs.
Role type
Senior IC AI safety researcher (evaluations)
Builds
Novel evaluation benchmarks and taxonomies for frontier AI models
Domain
AI safety and security research
Deliverable
production ML models
Required skills
AI safety evaluation, taxonomy design, distributed inference, codebase maintenance, researcher direction, research publication, English fluency
Preferred skills
Post-training (SFT/DPO/GRPO), agentic evaluation, conference presentation, travel
Technologies
vLLM, distributed inference frameworks
Responsibilities
Ship benchmarks on a 2-3 week cadence, maintain quality bar and rubrics, manage evaluation process and timelines, set quarterly roadmap with research leads, engage with AI lab ecosystem and conferences
