Research Engineer: Model Evaluations & Metrics

Anthropic; United States; Remote; USD 120,000 - 230,000; Full time

Job summary

Anthropic is seeking Research Engineers to design evaluations that quantify Claude's capabilities, reasoning, safety properties, and alignment with leadership expectations.

You will build scalable eval infrastructure, run experiments across live checkpoints, and present clear results to researchers and decision-makers to push Anthropic toward leadership in well-characterized AI systems.

Qualifications

• Strong Python programming skills, production or research infra experience.
• Experience building or operating distributed systems and data pipelines.
• Clear written and verbal communication to explain technical results to non-specialists or leadership.

Responsibilities

• Design and run evaluations of Claude's capabilities, safety, and behavior.
• Build and harden a scalable evaluation platform used during training and testing.
• Own dashboards and visualizations to monitor model health and evaluation outcomes.
• Debug anomalous eval results during training and communicate findings clearly under time pressure.
• Collaborate with research teams across the full lifecycle of new capabilities, from measurement to interpretation.

Skills

Python

Distributed Systems

Communication

Job description

Anthropic is seeking Research Engineers to design evaluations that quantify Claude's capabilities, reasoning, safety properties, and alignment with leadership expectations.

You will build scalable eval infrastructure, run experiments across live checkpoints, and present clear results to researchers and decision-makers to push Anthropic toward leadership in well-characterized AI systems.