Product Engineer, UK at Partly in London

Job description

Note: Partly is headquartered in Austin, TX with offices in London, UK, Christchurch, NZ and Auckland, NZ. Wherever you're based, we'll connect you with your nearest office for onboarding, and fly you to join the full team for our quarterly "Season Openers" (we cover travel and accommodation). If you're relocating to join us, we can also assist with relocation costs. 🚀 Our Story Partly is building the infrastructure for anyone to repair anything. We believe everything built should be repairable, and we're building frontier AI to put repair knowledge to work, starting with the $2tn automotive market. Our model, Interpreter, is the first AI purpose-built to understand repair. Trained on 1M+ car models and 1B+ parts, it reads a repair job the way an estimator does, resolves damage into the exact purchasable part, and shows its reasoning with a confidence score. On real production jobs, an estimator working with Interpreter reaches 98.8% parts-list accuracy, against 92.1% with today's tools. It powers PartlyRepair, used by collision shops, workshops and suppliers to take the admin off the job; PartlyLabs, the agentic infrastructure enterprises use to deploy Interpreter across their own workflows; and a developer API. Founded by ex-Rocket Lab engineers, we've raised a $50m Series B led by DST Global (Anthropic, Airbnb, Meta, TikTok, Spotify), with Blackbird Ventures, WNDR, Activant Capital, Icehouse Ventures, Square Peg, Airtree and Ecliptic Venture Capital. Hear it from the team: Why Partly exists and AI at Partly. 🖍️ This role This is a product-builder role, not a ticket-taking one. You own a product surface end to end: you talk to the customer, form the idea, build it, ship it, measure it, and decide what happens next. The same person who understands the repairer writes the code. There is no handoff for product quality to leak through. You'll build at the center of a multi-trillion-dollar industry that still runs on phone calls, spreadsheets, and tribal knowledge. Replacement parts is one of the largest physical markets on earth and one of the least digitised, and there is no established playbook to copy. Partly is creating the category. Our moat is Interpreter, a domain-specific AI model that understands parts, vehicles, and the messy real-world relationships between them in ways general-purpose models cannot. It is what lets software finally make sense of a physical, offline industry, and it is the engine underneath everything you build. And what you build is changing shape. We are not bolting AI onto old screens. We are creating genuinely new agentic experiences for a broad spectrum of users, from estimators and parts managers to accountants and suppliers, where people express intent and supervise outcomes while agents handle the mechanical work. Designing how that trust is earned, surface by surface, while each one still delivers value today, is the heart of the role. You'll work at an inflection point in how software itself gets built. AI-assisted development has collapsed the distance between an idea and a working version of it. The builders we want already work this way: they prototype in an afternoon, put it in front of real customers within days, and read the signal themselves. We want someone who works like that now, or is visibly building toward it. 💻 What will you do • Build, not just specify. Prototype new product iterations yourself using AI-assisted and LLM-assisted tooling, fast, disposable, real. Treat the prototype as the hypothesis and put it in front of real customers before writing the spec. Ship production code, instrument it, and watch how it actually gets used. • Stay close to customers. Spend time where the work happens. Understand what estimators, parts managers, and others actually do, where they lose time, and what better would feel like. This is a core weekly activity, not an occasional input. • Reconcile signal. Combine what you see with what the data says, and reconcile the two when they disagre