Raydar
Research Scientist, AI Safety
$250K–$450K
San Francisco, California · Full-time · Posted Oct 31, 2026
- Experience
- Hands-on safety work on large-scale models within roughly the past two years; no total years specified.
About the job
Design and run safety evaluations, benchmarks, and test environments for advanced AI models, and share reproducible findings with engineers and external research partners.
What you'll do
- Create standardized tests and environments that probe model capabilities, safety failures, and training improvement.
- Review and run evaluations, scrutinizing test quality and tracking model behavior and capabilities over time.
- Work with engineers and partners to document, reproduce, and communicate evaluation results.
What they ask for
- Hands-on experience with post-training and reinforcement learning.
- Hands-on safety work on large-scale models completed within roughly the past two years.
- Working knowledge of frontier model capabilities and the direction of the field.
- Peer-reviewed machine learning conference publications or established evaluation datasets.
Nice to have
- Experience evaluating models on AI R&D tasks, such as machine learning research or model training.
- Prior work with leading AI labs or government AI safety bodies, including pre-deployment testing.
Pay
USD 250,000 to 450,000 per year (base salary)
AI focus
Research and develop evaluations of advanced AI model capabilities and safety, including model performance on machine learning work and self-improvement of training.
Model-written summary · the original posting is authoritative