Research Engineer, Benchmarks

Join a small, technical team focused on AI evaluation to design and build rigorous benchmarks for frontier AI agents. You will translate real-world workflows into evaluation tasks, build scalable infrastructure, and develop metrics to assess agent performance, ensuring credible insights into their capabilities.

• Design and implement AI agent benchmarks for domain-specific tasks
• Build infrastructure for large-scale model and agent evaluation
• Develop metrics and analyses for benchmark reliability and difficulty
• Collaborate with subject-matter experts to model real workflows
• 2+ years experience in AI benchmarking or evaluation infrastructure
• Proficiency in Python, Docker, and Linux