AI Engineering Intern - Self-Healing Software, London

Light Inc; Greater London; On-site; GBP 24,000 - 32,000; Full time

Benefits offered by this job

Internship compensation above market

London office in-office role

Ownership from day one

Job summary

Light is building organic software and redefining how software is built and delivered. As an AI Engineering Intern you’ll work on agents, models, and infrastructure that turn organic software into something customers can rely on, with access to our repositories and internal coding agent.

This role is based in our London office. You’ll own projects from day one, ship work into production, and contribute to measurable improvements in model accuracy and reliability.

Qualifications

• Strong programming ability and practical AI tooling experience.
• Proven project work beyond coursework or tutorials: model, agent, tool, or application with tangible results.
• Clear thinking and ability to explain decisions, what you tried and why it didn’t work.

Responsibilities

• Help agents learn from their work by reviewing merged PRs and corrections.
• Train and deploy LLM and ML models; run experiments and compare results.
• Build and own end-to-end agents and their interaction with customers and production systems.
• Develop evaluations and tests to catch failures before they impact users.
• Deliver projects from scoping to deployment with measurable impact.

Skills

Strong programming ability

Hands-on AI tooling

Explain your decisions

Education

University of Oxford or University of Cambridge

Tools

LLM APIs

Agent frameworks

Job description

Light is building organic software and redefining how software is built and delivered. As an AI Engineering Intern you’ll work on agents, models, and infrastructure that turn organic software into something customers can rely on, with access to our repositories and internal coding agent.

This role is based in our London office. You’ll own projects from day one, ship work into production, and contribute to measurable improvements in model accuracy and reliability.