Software Engineer, Frontier AI Safety & Research

Aisi; Greater London; Hybrid; GBP 65,000 - 145,000; Full time

Benefits offered by this job

Talented colleagues

Direct influence on AI governance

Shape public-interest AI research

Pre-release access to frontier models

Pension contributions

Job summary

AI Security Institute's Research Unit is seeking software engineers to join the Core Technology Team in London. You’ll work on critical research software and infrastructure with mentorship from senior staff and interaction with world-renowned researchers from top labs.

You’ll contribute to projects like Inspect, model hosting, and evaluation systems, while collaborating with a leading team on open-source tooling and frontier AI safety initiatives.

Qualifications

• Experience writing production quality code.
• Ability to design, ship and maintain complex tech products.
• Strong written and verbal communication skills.
• Experience working with a world-class research team.
• Proficiency in Python and ecosystems.
• Experience building systems on AWS or other clouds using IaC.

Responsibilities

• Add a feature to Inspect sandbox plugins for advanced agentic evals.
• Support hosting/open-weight model classes on our platform.
• Assist research teams with custom infrastructure for new projects.
• Collaborate with open-source community on Inspect plugin features.
• Debug API changes in lab SDK for evaluation tests.

Skills

Python programming

Written communication

Verbal communication

Teamwork

Interpersonal skills

Production code quality

Tools

AWS

Terraform

Docker

Kubernetes

Job description

AI Security Institute's Research Unit is seeking software engineers to join the Core Technology Team in London. You’ll work on critical research software and infrastructure with mentorship from senior staff and interaction with world-renowned researchers from top labs.

You’ll contribute to projects like Inspect, model hosting, and evaluation systems, while collaborating with a leading team on open-source tooling and frontier AI safety initiatives.