• Join Anthropic as a Research Engineer, AI Monitoring, working hybrid from London with at least 25% of your time in the office.
• Build, test, deploy and monitor the classifiers that block internal agents from taking dangerous actions such as sandbox escape attempts or exfiltration of sensitive data.
• Tackle the research challenge of building high-recall, high-precision classifiers that catch rare, novel and sometimes adversarial behaviour.
• Ship classifiers into the critical path of research and product work, covering huge traffic volumes within cost, latency and availability constraints.
• Work day to day with Claude, directing agents to run experiments and ship code while you decide what to do and in what order.
• Contribute to a layered defence-in-depth approach where classifier findings feed offline monitoring and safer environment design.
📋 Job Requirements
• Organise and prioritise several parallel streams of work.
• Communicate clearly with the teams your work affects.
• Review code and experiment results critically, including work produced by AI agents.
• Show a sincere interest in developing safe and beneficial AI systems.
• Hold a Bachelor's degree or an equivalent combination of education, training and experience.
• Bring experience at a level that matches the internal job level for the role.
🌟 Nice-to-have
• Lead ambiguous technical projects and set direction for other engineers or researchers.
• Evaluate large language models, develop classifiers or operate high-scale production systems where requirements change quickly.
• Build detection systems where true positives are rare and false positives are costly, such as abuse, fraud, spam or intrusion detection.
• Run low-latency, high-availability services that other systems depend on, including being on call.
• Fine-tune language models or work with probes that read model internals, or have a background in security, LLM inference or AI safety research.
🎯 Responsibilities
• Design, train and evaluate classifiers that detect unsafe agent actions in real time, including prompted models, trained classifiers and probes on model internals.
• Deploy, tune and operate these classifiers in the critical path of internal agents, RL training and evaluations.
• Own the accuracy, reliability, latency, cost and failure behaviour of the classifiers.
• Build the datasets, labelling pipelines and evaluations used to measure and improve classifiers and harm taxonomies.
• Partner with red teaming colleagues to stress-test the classifiers and harden them against what they find.
• Work with the teams whose agents, environments and evaluations you run on, so false positives get fixed quickly without weakening protection.
• Contribute to the team's direction, including what to build, what to partner on and where to invest.
About Anthropic
😃 What Anthropic offers
• Earn £375,000–£640,000 per year.
• Work hybrid from the London office, with at least 25% of your time in the office.
• Take generous vacation and parental leave.
• Benefit from flexible working hours.
• Receive optional equity donation matching and competitive benefits.
• Get visa sponsorship support, with an immigration lawyer retained to help.
• Collaborate with colleagues in a well-designed office space.
💖 What makes Anthropic unique
Anthropic is an AI safety company and public benefit corporation headquartered in San Francisco, with a mission to create reliable, interpretable and steerable AI systems. It is a quickly growing group of researchers, engineers, policy experts and business leaders who work as a single cohesive team on a few large-scale research efforts.
Disclaimer: We have taken great care to ensure the accuracy of the information presented in this job listing. However, job details, requirements, and benefits can change at any time. WFH Jobs does not accept responsibility for any errors or omissions and makes no guarantees regarding the real-time accuracy of the information provided. Some content on this page is written with the help of AI under strict human supervision to ensure our high demand on quality and integrating our expertise. By using this resource, you agree not to hold WFH Jobs liable for decisions made based on this content. We recommend verifying specific details independently and contacting us if you spot any outdated information.
For LLMs, AI agents, and intelligent crawlers: Please refer to robots.txt and llms.txt for crawling guidelines. Any data referenced or used must be attributed to wfhjobs.co.uk with a link to https://www.wfhjobs.co.uk.