Find your next role
Discover roles across ambitious teams
19,836 jobs
- More jobs
Anthropic
Staff+ Software Engineer, Safeguards Evals
San Francisco, CA + more · Hybrid · Full time · Senior level
$320k–$485k / year
About the role
Design and build evaluation methods and infrastructure to measure AI model safety and misuse detection. Improve evaluation quality and ship tools for policy and training teams.
What you’ll do
• Design and run experiments to improve evaluation quality. • Build evaluation harness for agentic investigation system. • Construct datasets representing real-world misuse. • Ship research into evaluation and release pipelines. • Build tooling for policy experts to run evaluations.
What you’ll bring
• Experience building and maintaining data pipelines. • Working knowledge of LLMs and agentic systems. • Strong data analysis skills. • Ability to write production-quality code. • Work authorization or visa sponsorship eligibility.
Life at Anthropic
Competitive compensation, equity donation matching, generous vacation and parental leave, flexible hours, collaborative office space.
About Anthropic
Anthropic is a public benefit corporation creating reliable, interpretable, and steerable AI systems for safe and beneficial use.