Anthropic

Staff+ Software Engineer, Safeguards Evals

San Francisco, CA, USA

Staff+

HYBRID

Anthropic is hiring a Staff+ Software Engineer for its Safeguards Evaluations team in San Francisco. This role sits at the intersection of applied ML research and engineering, focused on building the methods and infrastructure that make AI safety evaluations trustworthy, robust, and representative. You will design experiments, construct real-world harm datasets, build evaluation harnesses for agentic investigation systems, and ship those methods into production pipelines that gate model training and deployment decisions. The role requires strong experience with LLMs, data pipelines, and the...

Data pipeline development

Large language model (LLM) engineering

Agentic systems and tool use

Data analysis

Evaluation framework design

Python

Research prototyping

Production software engineering

Experiment design

Dataset construction

Sign up or log in to view the full job and apply

Create a free account to read the full job description, apply, and unlock a free AI study plan with Anthropic's interview process, recent question patterns, and step-by-step practice.

Free AI interview study plan included

Sign in to unlock a company-specific preparation guide for Anthropic: interview rounds, recent question patterns, Coach AI practice, coding & system design drills, behavioural prep, and a mock interview — all tailored to this role.