Build AI
Member of Technical Staff, Evals Lead
San Francisco · On-site · Staff · Full-time · Posted Aug 30, 2026
About the job
Lead rigorous benchmarks of video and world-model capabilities to determine how Build AI’s data affects model performance and inform research, training, and data decisions.
What you'll do
- Design and run comparable benchmarks with and without Build AI data, including held-out tasks and contamination and leakage checks.
- Build analytics to track performance, failure patterns, and how dataset scaling affects capabilities.
- Coordinate with research and dataset-quality teams to improve evaluation and data-collection processes.
What they ask for
- Have shipped or driven consequential evaluations that changed training, hiring, product, or data decisions.
- Understand evaluation philosophy, including validity and what results can legitimately claim.
Nice to have
- Experience with video, robotics, world models, or multimodal evaluations.
- Experience designing evaluations for papers, product launches, or data decisions, or with evaluation operations.
AI focus
Evaluate video and world-model capabilities, measure the effects of training data, and develop repeatable evaluation processes.
Model-written summary · the original posting is authoritative