AI Product Engineer (m/f/d)

We make healthcare easy.

And we have already built the hard part. Around 6,000 patient bookings run through haelsi every month, with roughly 700 new patients joining on top of that, across 3 health centers in Vienna, online, and 100+ companies, with over 150 providers in our team. We are growing strong double digits year over year, and we are just getting started. Our patients love us: we constantly hit an NPS over 80, basically unheard of in healthcare. With haelsi weight, our fresh out of the box GLP-1 online program, we already have hundreds of happy patients and are growing fast. More telemedicine lines are coming.

Now we are building Austria's first AI-native healthcare company. Not AI bolted onto old workflows. AI in the bones, humans where it counts.

Operating philosophy: AI-first, human touch. AI takes the repeatable work so our people can spend their time where only humans shine. The goal: better outcomes at lower cost. We are adding builder power to the product team so we can ship more products that make healthcare easier for our patients.

How we work

We are lean and we keep it that way on purpose. No rigid structures, no process that thinks for you, no layers between you and the work.

• Big influence. In a small team your work decides whether a whole product wins. More ownership, more impact, fewer rules.
• Everyone is a builder. You pull work from a prioritized backlog and own it end to end, from the first stakeholder conversation to shipping, frontend to backend. You bring your own ideas, take the tickets that play to your strengths, and get help on the rest. When you need design muscle, you spar with our experts. We are devs with strong product taste, or PMs who vibecode and design. The title matters less than what you can ship. No one assigns you work, no one babysits.
• AI-first for real (yes, we are AI-pilled). Claude, Claude Code, and Claude Design are part of the daily workflow, not a side experiment (ofc we also use other LLMs). Claude automatically writes and prioritises tickets against our prio rulebook and checks that we do it right. We made it our super brain PM. Every meeting gets a transcript, agents pull the to-dos out of it and pre-write the tickets for the team.

What you'll do

• You bring agentic AI expertise into the product team and point it wherever it creates the most value. Today that is mostly haelsi weight, our GLP1-Telemedicine program and a new AI booking agent, but the focus will move to whatever moves the needle next.
• You design, build, and ship AI agents that do real work: agents that guide and support patients, triage and route, or run a multi-step flow end to end, and conversational agents that use tools and render results as interactive UI instead of a wall of text.
• You own the agent craft end to end: prompting, tool-use and function-calling, retrieval over our data, plus the evals, guardrails, and monitoring that keep agents reliable and safe in production. Setting them up, training them, and maintaining them is the job.
• Where agents touch clinical topics, you keep them on the right side of the medical-device firewall: they surface information, context, and drafts for a clinician, they never make the clinical call (We are not building medical products).
• You are still full stack. The agents run through our chat channel and event bus and ship as real product, TypeScript across the board with Nuxt and Vue, Strapi, Supabase, Stripe, and vibecode tools like lovable.

Your bag of tricks

• Commercial mindset. You are not a code enjoyer. You light up when a product ships, gets used by a lot of people, and is loved. Code is a means to an end, and the end is impact.
• Strong product taste. You have opinions on what should exist and why. You can say "wrong workflow" and back it up. As AI commoditizes writing code, taste is the moat and judgment is the job, not a bonus.
• You build agents, not just prompts (you are AI-pilled). You have shipped LLM agents in production: tool-use and function-calling, retrieval over real data, structured outputs, multi-step orchestration. Roughly two to three years building products plus about a year of hands-on agent work.
• You keep agents alive. You write evals, set guardrails, and monitor agents in production so they stay reliable and safe as they scale. Training and maintaining them is the job, not a one-off launch.
• Builder energy. You have shipped your own things. Indie hacker, side projects, a graveyard of repos and a couple that actually went live, the entire internal stack for a small company once. Scrappy beats polished.
• System thinking. You see how the pieces fit and design data and workflows that hold up, instead of leaving a mess for the next person. You value properly set up systems, and you are lazy in the right way: you let AI do whatever it can do well, wherever it fits. It is a balance.
• No ego, only outcomes. Titles and hierarchy are flat here, and there is no room for ego, only for what we ship together. You flex into what the team needs and care about making healthcare easy more than about any single feature or title. When work gets reprioritized or killed, your energy holds.

We are not after the most senior engineer or a clean-code purist. Taste, range, AI fluency, and drive beat years on the CV. Short version: we are nerds hiring nerds. Builder-business nerds with ideas, drive, and the skill to properly go off in the age of AI.

Nice to have

• English is our working language. German is a bonus.
• You have built with an agent framework or generative UI (Claude tool-use and MCP, LangGraph, the Vercel AI SDK, or similar).
• Background in healthcare or another regulated, GDPR-heavy environment.
• Solo founder or indie hacker past, with shipped projects you can show.
• You already know parts of our stack: Nuxt and Vue, Strapi, Supabase, Stripe, lovable.

What we offer

• Direct work with the CPO and the whole company, where what you ship actually moves the business
• Fully remote or hybrid, your call. We have remote people on the team and still love getting together at our Vienna office in the city center (1080, Palais Auersperg)
• Market-rate salary. By law we have to state the collective-agreement minimum (€2,059.41 per month, full-time), but that is only the mandatory floor, not what you will actually earn here. We pay in line with the market, based on your qualification and experience.
• MacBook Pro
• Full Claude API credits, Claude Code, Gemini, and any other AI tool you need in your workflow
• Klimaticket Wien, access to all haelsi healthcare services for you and your loved ones, and 25 vacation days plus the usual Austrian public holidays
• A team that works fast and honestly, and has fun doing it

How we hire

Lean and fast, no endless loops. Usually three steps:

1. Intro call with Christopher, our CPO
2. Tech deep-dive with our Tech Lead and UX Lead
3. Final chat on fit and growth with the team

Sometimes we add a small take-home challenge so you can show how you actually build. We move quickly and keep you in the loop.