Anthropic
Research Engineer, Post-Training Model Evaluations
$500K–$850K
San Francisco, CA +2 · Full-time · Posted Apr 28, 2026
- Location
- San Francisco, CA
Seattle, WA
San Francisco, California, United States - Experience
- Years of experience required will correlate with the internal job level requirements for the position.
- Education
- Bachelor’s degree or an equivalent combination of education, training, and/or experience; required field of study is a field relevant to the role as demonstrated through coursework, training, or professional experience.
- Visa sponsorship
- Visa sponsorship is available, but not for every role and candidate; Anthropic says it will make every reasonable effort to obtain a visa for candidates it offers a position and provides immigration lawyer support.
About the job
Leads evaluation strategy and live monitoring for production Claude post-training runs, using trustworthy measurements to assess model quality and guide training and release decisions.
What you'll do
- Optimize the production evaluation mix, monitor evaluations during training, and advise teams on evaluation methodology.
- Investigate regressions and inform training interventions; develop dashboards, alerts, and reports for tracking model quality.
- Help evaluation authors meet production standards and provide clear recommendations based on results.
What they ask for
- Experience designing, running, and analyzing evaluations for ML models at scale.
- Strong Python skills and comfort working with production systems.
- Ability to interpret ambiguous results, influence teams, and debug complex, time-sensitive issues.
Nice to have
- Hands-on experience post-training large language models; research experience in ML evaluation or benchmarking.
- Background in statistics and experimental design; experience developing robust evaluation metrics for ML systems.
Pay
$500,000 — $850,000 USD (base salary)
Tools
Python
AI focus
Evaluates production Claude models during post-training to assess model quality, capabilities, alignment, and safety, and to inform training and release decisions.
Model-written summary · the original posting is authoritative