Description du poste
Leadership: 6+ years of experience, including leading a small team Other people’s systems: you’re comfortable with asynchronous pipelines, legacy interfaces and code written by others. Hospital experience isn’t required Evals: you can answer “how did you know it was good enough?” with ground truth, regression tests and a clear taxonomy of error classes Nice to have: German, as the catalogues, the coding rules and much of the work with our experts are in German; an ML background, including judgement on when fine-tuning pays off; retrieval and source attribution; healthcare data standards such as FHIR; regulation such as MDR and the EU AI Act Hands-on with LLMs: you’ve built with LLMs yourself and understand where they fail: non-determinism, cost, latency and evaluation. You can design systems around them and challenge our AI engineers’ choices. Experience running an LLM or agent system in production is a strong plus, and strong personal projects count Communication: you explain decisions and trade-offs clearly in English in a remote team Full-stack depth: you go deep across backend, frontend and data, and you commit code
German hospitals are paid a fixed amount per case, based on the codes assigned to it. Insurers dispute a share of the bills through an external medical audit, and the hospital can appeal when the audit cuts a bill. Recare is building medical coding agents to rebuild this chain: they read the case file, propose codes, assess audit findings and draft appeals. A subject matter expert reviews every result You’ll lead the technical development of this product. The role is about 80% hands-on building and 20% leading, and it grows with the platform into areas beyond billing Technical direction: set the architecture with the team across backend, frontend and a data model built on healthcare standards such as FHIR. Set the technical roadmap with the team and deliver on it predictably Agent orchestration: design how models, deterministic rules and human review fit together. Work with our AI engineers on system design Hands-on delivery: build the backend for cases, documents and long-running agent runs, including the append-only decision log that keeps every case audit-proof when an insurer challenges it months later Product partnership: be the technical counterpart to the product manager and the subject matter experts on planning, specs and trade-offs Production quality: own reliability, observability and security in a sensitive healthcare context, with an eval harness built from expert-reviewed cases as a CI gate AI coding agents: engineers use them every day and take responsibility for the code we ship. Product managers and designers prototype in our codebase too, and engineers review what goes to production Human-in-the-loop: agents propose, the expert accepts, corrects or rejects, and every decision is logged with its evidence Shipping: we release early to development partners and switch features on per hospital Data protection: EU-only inference and no patient data in logs are hard constraints Ownership: product owns what we build and why. The tech lead owns how it’s built and that it ships Stack: TypeScript, Node.js and Go, Python for inference, Postgres, React, AWS Bedrock, Docker and Datadog Success: measured in revenue hospitals keep, not in usage Benefits Flexibility in how you work: Work is not measured by hours or location. Work remotely, come to the office, or spend time working from abroad. Staying connected: We stay in touch through clear communication and the right tools. Work is transparent, and teams stay aligned without unnecessary overhead. Trust and space to focus: We trust people to manage their work and take breaks when needed. What matters is staying focused and moving things forward. Work from abroad: Spend time working abroad, up to 182 days per year. We support flexible setups beyond your usual location. Office space in Berlin: A central place in Berlin Mitte where people come together to meet and collaborate. Join whenever you feel like it. Learning & Development: Grow through regular feedback, clear development paths, and a dedicated learning budget. Your birthday, full off: Take an extra day off to celebrate yourself. No Slack, no calls, no work. Tools, hardware & AI: We equip you with a MacBook Pro, modern collaboration tools, and AI that is part of how you work every day. Monthly Edenred card: Enjoy a €50 monthly budget on your Edenred card, flexible to use online or in-store. Choose your own setup: Pick the tech and equipment you prefer. We support you with a dedicated budget. Time together: Regular opportunities to connect and spend time together, from team meetups to company-wide events. A strong team around you: Work with people who take ownership, support each other, and enjoy working together.