Software Engineer Intern

Software Engineering InternshipLightWork AI · London · Six-month internship · 4 days a week in the office · Immediate start About LightWork AI LightWork AI is building the operating system for the property industry. We have built an agentic AI assistant that automates the operational workload that slows down letting and estate agents. Felicity manages communications, compliance workflows, maintenance coordination and scheduling, enabling property businesses to operate with the efficiency of modern digital platforms. The role This is a six-month internship for someone who wants a career in engineering. You will learn the system the way the best engineers do: by testing it, breaking it, and working out where and why it falls short of what a customer expects. In a company built on agentic AI, trust comes from quality. Felicity does not draft a reply for someone else to send. She reasons, calls tools and takes action in live customer systems, so the same enquiry can be handled two different ways and no unit test will tell you whether the product is good. You will work alongside our engineers on how we know Felicity is performing. We are a small, AI-augmented team: our engineers work alongside Claude Code and AI agents every day, and much of our weekly rhythm runs through agentic workflows. You will be part of that from week one. There is room to grow. In time, and provided you are performing well across all elements of the role, there are opportunities to convert into a permanent seat. What you'll do ● Get inside the agentic pipelines behind Felicity: prompts, tool calls, retrieval and state. ● Write and maintain test scenarios across product areas such as prospecting, maintenance, compliance, email and voice, and track how the product performs against our evaluation suites. ● Take part in the daily review of flagged conversations: work out what each one means, separate configuration, data and user error from genuine bugs, and log the outcome. ● Debug systems that are not deterministic, where the answer to “why did it do that” is a chain of decisions and tool calls rather than a stack trace. ● Write real code: test harnesses, fixtures and internal tools, reviewed by the engineer who owns that area. ● Work across our stack as you go: Python and FastAPI, LangChain and LangGraph, our NestJS microservices, MongoDB, Qdrant and Redis on Google Cloud, with CI/CD through GitHub Actions. Who you are 1. Hands-on with AI. You use agentic coding tools as a matter of course rather than as novelties, and can show us what you have built with them. Some exposure to LLM APIs, prompt engineering or vector databases, from a job, a course or your own projects. 2. A builder. Things you have built yourself: a side project, an agentic workflow, or something you automated because it annoyed you. Expect to talk one through and say which parts were yours, not the model's. 3. Methodical about root cause. You are not satisfied with what broke. You want to know why, and you keep going until you can explain it. 4. Clear in writing. A bug report or a test case is only as useful as it is readable. 5. Ready for the pace. We are a lean team and things move quickly, so you will need to be comfortable picking work up and seeing it through. Previous engineering experience, TypeScript, Git in a team, or exposure to testing frameworks such as Pytest or Playwright are all welcome, but none is required. The deal This is a six-month internship, based in our Victoria office four days a week with one day remote. You will need the right to work in the UK, as we cannot offer visa sponsorship, and we are looking for a first-class or strong degree in Computer Science, Engineering or Mathematics, graduated within the last 18 months. Compensation ● Competitive salary ● Free dinner if you stay after 7:30pm