AI Safety Research Residency — Remote UK

PromptCube; Greater London; Hybrid; GBP 265,000 - 375,000; Full time

Benefits offered by this job

Mentorship from Anthropic researchers

Access to Claude models and GPUcompute

Remote UK placement with London office

Job summary

Anthropic Residency is a paid, fixed-term program placed remotely in the UK with an option to work from the London office. Placements run six weeks to six months, pro-rated for duration, with mentorship from Anthropic researchers and access to Claude models and GPU compute.

The program emphasizes transitioning experienced technologists into empirical AI safety research. Applications close 22 November 2026, with offers expected in January 2027.

Qualifications

• Motivation to make sure AI is safe and beneficial for society as a whole
• A record of leading ambitious work in a variety of technical fields
• Ability to thrive in fast-paced, collaborative environments
• Background in machine learning is not required
• Experience authoring research papers in ML safety or NLP (preferred)

Responsibilities

• Work full-time on a concrete, scoped project with direct mentorship from Anthropic researchers
• Carry out empirical research in AI safety, alignment, scalable oversight and robustness
• Use Claude models and GPU compute to advance the project and connect with the AI safety community

Skills

Motivation to keep AI safe and benefit”

Leadership in technical fields

Collaborative environment

Background in machine learning not req

Experience with ML safety & alignment

Ability to communicate clearly

Tools

Kubernetes

LLMs

Reinforcement learning

Job description

Anthropic Residency is a paid, fixed-term program placed remotely in the UK with an option to work from the London office. Placements run six weeks to six months, pro-rated for duration, with mentorship from Anthropic researchers and access to Claude models and GPU compute.

The program emphasizes transitioning experienced technologists into empirical AI safety research. Applications close 22 November 2026, with offers expected in January 2027.