AI engineer

Job description

Making preventive health the default for every human in Europe The mission At Lucis https://lucis.life/?utm_source=career_job_desc&utm_medium=content, we believe healthcare should be preventive, not reactive. We’re building the OS for human longevity to help people add more healthy years to their lives. You can read more about who we are and how we work here https://lucis.life/careers?utm_source=career_job_desc&utm_medium=content. THE ROLE We're building AI agents that act as a doctor for preventive health — analyzing 110+ biomarkers, reasoning over clinical data, and delivering protocols that change how people age. The system works. Now we need to make it smarter, faster, and ready for millions of users. As our AI Engineer, you own the reasoning engine: the agents, the retrieval pipelines, the clinical workflows, and the infrastructure that makes it all reliable in production. You work directly with our medical team to translate clinical knowledge into systems that actually behave like they understand it. This isn't a research role. It's a build role. The gap between a great LLM demo and a trustworthy AI doctor is enormous — your job is to close it. You could be responsible for - AI agent architecture — design and ship the multi-agent system behind our AI doctor: reasoning, planning, tool use, clinical citation, and safe fallback behavior - Retrieval & knowledge — vector infrastructure, embedding pipelines, RAG architecture over clinical literature and user health data. You know the difference between retrieval that works and retrieval that a clinician would trust - Data pipelines — ingestion and normalization from labs (Eurofins, Randox), wearables, and third-party partners. Clean data in, reliable inference out - Production reliability — LLM endpoints that reason correctly at scale, with observability, evals, and failure modes you've thought through before they happen - Medical collaboration — translate complex clinical requirements into agent workflows, working side by side with our medical advisors. You don't need to be a doctor, but you need to earn their trust ABOUT YOU You’re passionate about the future of human health and want your work to help people stay healthy for longer. You move quickly from idea to execution, take full ownership of what you build, and work best with talented people who care as much as you do. You thrive in fast‑moving environments, learn by doing, and value feedback as a way to continuously improve. You’ll fit in well if - You have 8+ years' experience building and shipping software and ML systems in production - You've shipped AI agents in production recently — not as a side project, as the core product - You think in systems: prompts, retrieval, tool orchestration, evals, and failure modes are all part of the same design problem - You're hands-on with vector databases, retrieval pipelines, and LLM endpoints — Python-native, comfortable in LangChain or equivalent - You write evals before you ship, because you know that vibes-based QA doesn't work for clinical reasoning - You've worked with messy real-world data (health, finance, legal) and built pipelines that handle it without breaking silently - You're genuinely obsessed with health — you track your own biomarkers, read PubMed, or are just deeply frustrated that healthcare is still reactive We might not be a fit if - You need a clearly defined role with stable responsibilities. - You prefer strategic advisory work over hands-on execution. - You've only worked in large, well-established companies. - You need perfect information before making decisions. - You prioritise predictable 9–5 work over mission intensity. - You're uncomfortable with frequent context-switching and urgent pivots. Our current stack Python repo and TS monorepo (platform) · PostgreSQL + Prisma · AWS (Terraform) · GitHub Actions · Claude Code + Cursor · AI/agent framework: mostly Langchain ecosystem THE PROCESS - Intro call (20 min): culture & role fit. - Technical interview (30 min) - At-home case study: hands-on project, delivery in 2 days. - Deep dive (90 min) and team chat (on-site) - Reference calls HOW WE WORK We work together from our Paris hub. We’re passionate about what we’re building and believe the fastest way to create something exceptional is side by side. We’re open to relocation support for the right individuals, and we welcome missionaries who travel to work with us in Paris on a regular basis. - Rigor without ego: Audits, science, and code all deserve the same high bar. - Radical ownership: Feedback loops are short; everyone contributes to building the best version of Lucis. - Velocity over perfection: We ship daily and prefer a good decision today over a perfect one next week. At Lucis, AI isn't just our product, it's our engine. 100% of our teams are equipped with the best AI agents. You have carte blanche to explore and automate everything that can be, so you can focus exclusively on high-value work.