375K–640K GBP / year

Research Engineer, AI Monitoring

Machine learningHybrid — London, UK
Published on 2026-11-16
These details were extracted automatically from the original listing. They may not be complete or fully up to date — it's worth checking the original job post.

About this role

Build and operate classifiers that detect unsafe actions by AI agents across research, training, evaluations, and internal deployments. The role combines research into high-performing safety detection with production engineering for systems that must meet strict reliability, latency, and cost requirements.

About the company

Anthropic builds reliable and interpretable AI systems, including the Claude family of models, for businesses and developers. The company focuses on AI safety research and practical AI assistants for enterprise and consumer use.

What you'll do

  • Design, train, and evaluate classifiers for unsafe agent actions
  • Deploy and operate classifiers in critical paths, owning accuracy, reliability, latency, cost, and failure behavior
  • Build datasets, labeling pipelines, and evaluations for measuring classifier performance
  • Partner with red-team colleagues to stress-test and harden classifiers
  • Work with agent, environment, and evaluation teams to address false positives without weakening protections
  • Contribute to team priorities and technical direction

What we're looking for

  • Organize and prioritize multiple workstreams, and communicate clearly with affected teams
  • Critically review code and experiment results, including AI-generated work
  • Sincere interest in developing safe and beneficial AI systems

Nice to have

  • Experience leading ambiguous technical projects and setting direction
  • Experience with LLM evaluation, classifier development, or high-scale production systems
  • Experience building detection systems where positives are rare and false positives are costly
  • Experience operating low-latency, high-availability services, including on-call responsibility
  • Experience fine-tuning language models or using probes and other model-internal methods
  • Experience with LLM inference, serving, or training infrastructure
  • Security background, such as sandboxing or detection and response
  • Familiarity with AI control, scalable oversight, agent safety, or reinforcement learning

Benefits

  • Generous vacation and parental leave
  • Flexible working hours
  • Optional equity donation matching
View original job post