jobs

Research Engineer, Model Evaluations

Anthropic is seeking Research Engineers to build evaluations that measure what Claude can do across capabilities, behavior, and safety. The role focuses on turning ambiguous concepts such as intelligence, reasoning, agentic behavior, and mo

Company

Anthropic

Location

San Francisco or New York City, United States

Status

Closed

Posted

2026-09-13T03:31:18.256+00:00

Anthropic is hiring research engineers to build evaluations that measure Claude's capabilities, behavior and safety, turning vague ideas like reasoning and agentic behavior into usable metrics. The work covers designing and running evaluations, building distributed evaluation infrastructure that runs against training checkpoints, maintaining model-health dashboards, diagnosing anomalies during training runs, and improving tooling. The posting asks for strong Python plus experience building or operating distributed systems, data pipelines or research infrastructure, and values production support and on-call work. Salary is listed at 500,000–850,000 USD, in San Francisco or New York City.