Founding AI Engineer @handly - 20VC backed

Germany has about a million trade businesses. They do more than €700 billion of work a year, and they run almost all of it by hand: quotes in the evening, invoices on the weekend, payments chased between jobs. For decades, the software built for them was made for desk workers. Forms, menus, dropdowns, buttons. None of it was made for people who are always on the go and only care about the outcome. AI changes that. For the first time, the admin of a trade business can be done for the owner instead of by them. We are building a business-in-a-box for the trades. Tradespeople want outcomes done for them, with as little interaction with software as possible. handly takes care of leads, quotes, measurements, invoicing and payments, so they can focus on their craft, and is accountable for the result. We started with roofers in Germany. Other trades are next. Delivering outcomes is a much harder problem than delivering software. That is the job of an AI-native services company. Most of our vision has not been built yet, and we are looking for top-tier talent in Berlin to help us bring it to life. We are a small, close-knit team in Berlin-Mitte, backed by 20VC, Stride and Base10. We build, experiment and move fast, and we have found creative ways to use AI in every part of the company: product, ops, go-to-market and admin. about the role This role is for engineers who want to push what AI can do for a real business. As our founding AI engineer, you'll have high agency to hone your skills with AI and solve hard real-world problems that bring delightful experiences to our partners. The biggest opportunity at handly: we run the entire administration of our partners, rather than one vertical of their needs. That gives AI the room to take on large, autonomous tasks end to end, and gives us room to get very creative with what AI can do as we capture more and more channels of communication. You will own AI products end to end, make the core product and technical decisions, and turn messy, real-world workflows into AI that does the work for our partners. We are a small team working on: modernizing the user experience of an underserved industry reimagining workflows that have not changed in decades, with AI moving from AI workflows to high-autonomy agents capturing data and context from calls, emails, photos, documents and voice Tasks AI already runs in production across handly: intelligence on every customer call, automatic payment matching, a daily briefing for every partner. Some of the bigger projects you will contribute to or lead: Quote generation. Turn voice notes, photos and documents from customers and architects into a six-figure quote, priced intelligently from the partner's own catalog. Measurements. Measure buildings from satellite imagery, drone footage and public building data, without a site visit. Back office in natural language. An MCP layer over the whole product, so partners can manage their back office in a chat. Email intelligence. Capture intelligence from email threads and attachments, automate communication with clients and suppliers, and aggressive anti-phishing and spam protection. Self-improving pipelines. Use every correction, edit and rejection to improve the next output. Evals. Build the evaluation sets and benchmarks that gate every AI feature before it ships. Requirements Have 4+ years of full-stack engineering experience (exceptions apply for exceptional engineers) You have shipped AI features that real users depend on, and continiously evloved them. Strong full-stack TypeScript comfortable across backend, web and mobile. Experience building evals, benchmarks and self-improving loops that automate research on AI capabilities. Creative about capturing data and context (transcripts, photos, documents, voice) to improve AI features. Strong prompt engineering: structured outputs, tool use, agents, and good judgment about when an LLM is the wrong tool. Comfortable with ambiguity. You lead your proje