Research Engineer, Knowledge Foundations

About this role

Anthropic is seeking a Research Engineer to design and optimize training environments and evaluations for Claude's knowledge work capabilities. You'll conduct end-to-end experiments spanning data pipelines, model training, and evaluation to improve how Claude searches, retrieves, and reasons over information at scale.

What you'll do

• Design and iterate on training environments and data pipelines for knowledge-intensive reasoning tasks
• Run full-cycle experiments: form hypotheses, build infrastructure, train models, and analyze results
• Develop evaluations that measure progress on search, retrieval, and reasoning quality
• Identify model failure modes and translate them into training signals
• Collaborate with RL, post-training, and product teams to align priorities and ship improvements
• Build observability, dashboards, and operational tooling for training environments with high signal-to-noise metrics

What they're looking for

• Python engineering with production-grade, reliable code
• ML experiment design and analysis
• Full-stack ML work (data pipelines, training, evaluation)
• Large language model training, fine-tuning, or RL experience
• LLM evaluation design for open-ended domains
• Distributed systems operation at scale
• Clear communication across teams and time zones
• Comfort with ambiguity and prioritization