Senior Full Stack AI Product Engineer

Full Stack Ai (product) EngineerEngaige · Utrecht, The Netherlands · Full-time · Senior We are looking for a motivated and ambitious full stack (product) Ai engineer. Who is hands on and doesn’t have a 9-5 mentality. About EngaigeEngaige builds AI customer support agents for eCommerce. Our platform is built on insights from over 1 million support tickets, and it resolves customer conversations autonomously: policy-governed, transparent in its reasoning, and improving itself with every interaction. Our promise to customers is simple: keep your helpdesk, replace the AI. We are a team of 11 in Utrecht and brands running 120,000+ conversations a year through our agents. Small team, scaling and a roadmap that moves weekly. That is where you come in. The engineering team is small and relatively senior - people with profiles much like the one described here. About the roleWe are looking for an engineer who cares as much about what gets built as how it gets built. You will own features end to end: sit with the engineers in the morning focussing on future features, decide what the fix actually is, ship it that week, then look at the data to find out whether it worked. No handover to a product owner, no ticket thrown over a wall. On a small engineering team serving brands across Europe, someone who understands the customer and can write the code is worth several people who can only do one. This is not a role where the roadmap arrives finished. You will help write it. The problems you will work onThese are real, open, and on the roadmap right now. From 80% to 90%. Our agents resolve 60–80% of supported tickets at go-live. Getting a customer to 90% today still takes configuration work by their team. We believe the self-improvement loop should do that work instead. What does that product look like?Trust at a glance. Support leads need to see why an agent did what it did, and change its behaviour without calling us. Policies, guardrails, agent configuration: every one of those is a product surface.Measuring a non-deterministic product. Resolution rate, answer quality, escalation correctness, regression risk. We are building the evaluation harness that tells us whether last week’s change made the agents better or merely different.Depth in the integrations that matter. Shopify, Zendesk, Gorgias, Intercom, Freshdesk. Our agents do not just answer, they act: refunds, returns, tracking, subscriptions. Every new action is an integration problem and a trust problem at the same time.Ready for bigger customers. Our international deals are several times the size of our Dutch ones. That changes what the product has to do. What you will doOwn features end to end: the problem, the design decisions, the implementation, the rollout, and the metric that tells you it workedTalk to Sales and to our Customer Success team directly, then decide what that means for the productWork across the whole stack - the Python backend that runs the agents, the TypeScript dashboard our customers use every day, and the infrastructure underneath. We expect engineers here to be comfortable in all of it and to ship a change that touches every layer without needing a hand-offWork on agent behaviour with the team: policies, guardrails, evals, and the loop that makes the agents better every weekShip continuously. We optimise for the smallest piece that delivers value to a customer, and our processes are built so that we can release constantly while staying in full control of what goes outRaise the engineering bar: what we automate, what we review, and what we delete Our stackBackend: Python and Postgres, with Weaviate for vector storage and Redis as our memory store. We build event-driven, with Google Cloud Tasks as the delivery substrate. Frontend: TypeScript, React and TanStack Start. Infrastructure: GCP-first, with Terraform and GitHub Actions. AI: the Claude and OpenAI model APIs. Analytics: Cube on BigQuery. Day to day: Linear and Sentry. You do not need to have used all of it. Good engineers pick up a stack; product judgement takes longer to learn. We build with coding agentsWe build primarily with Claude Code, Cursor and Codex, and the whole company is organised around working that way - not just engineering. This is not an experiment we are running on the side; it is how the work gets done here. It also raises the bar rather than lowering it. We care a great deal about quality, technical quality very much included, and we deliberately keep humans in the design and review loop. An agent can produce a working change quickly; deciding whether it is the right change, and whether it is good enough to put in front of a customer, is still ours. Knowing when the output is ready to ship - and when it is not - is a large part of what we are hiring for. We expect you to get more out of these tools than we currently do, and to have opinions about where they should not be trusted. You might be a good fit if youHave 5+ years building and shipping software people actually use, ideally at a startup or somewhere with a genuinely high engineering barAre comfortable across the stack, and have shipped things where you owned both the backend and what the user seesHave a track record of end-to-end features with visible impact, not only refactors and incremental improvementsHave opinions about product, can defend them, and drop them when the data disagreesHave built with LLMs: prompts, tool use, retrieval, and the failure modes that come with themAre already native with agentic coding tools — Claude Code, Cursor, Codex or their equivalents are your core workflow today, not something you are planning to pick upWould rather ship and learn than wait for certaintyCommunicate crisply with engineers, with customers, and with colleagues who do not write code Even better if youHave worked on evals or quality measurement for a non-deterministic systemHave experience in customer service, CX, or eCommerce softwareHave been a founding engineer, or the first product-minded engineer on a teamHave strong visual and interaction taste, and can build a UI that feels fast and simple Your degree is not a hard requirement here. What you have built, and whether you can show it to us, tells us more. This role is not for you ifYou want fully specified tickets before you start building - deciding what the ticket should say is half the value of this roleYou want to go deep in one layer of the stack - that is a different job, and also a good one, just not this oneYou want fixed scope and a steady pace. Weeks here are uneven, and when a customer is blocked we fix it What we offer€78,000 – €95,000 gross per year (12 × monthly), plus 8% holiday allowance, depending on experienceA senior seat on an 11-person team with paying customers and revenue - not a pre-launch betReal ownership: the surface area per person here is larger than it will ever be at a bigger companyFlexibility that runs both ways. We ask you to stand up when it matters and you are able to, out of genuine interest in the work - and in return we are genuinely flexible with you, including a very flexible vacation policyBest-in-class AI tooling, no debates about seat countsWe work together in person from our office in the heart of Utrecht; specifics are open to discussion How to applyEmail work@letsengaige.com, or reach out to julian.bakker@letsengaige.com directly. To save us both a round trip, include: What you have built that you are proud of, with a link if there is oneYour GitHub, or anything else that shows the work itself A few honest sentences is plenty - it does not need to be an essay. We read every application and we reply to every one. If we take it further, come prepared to walk us through a couple of technical or product achievements you are proud of. We would rather talk them through with you than read a list.