Senior ML/AI Systems Engineer (MLOps & Product Reliability)

Stellenbeschreibung:

Aiomics is building a data and documentation intelligence layer for hospitals and clinics—deeply rooted in real frontline work with Europe's most advanced healthcare providers. Doctors, nurses, and therapists still lose hours every day to fragmented, incomplete information. We're fixing this by combining multimodal AI with structured validation loops that empower clinicians rather than overwhelm them.

We are a small, senior, impact-focused team with an unusually deep understanding of the problem space:

A medical doctor with a decade in digital innovation and years at the clinical frontline
A data scientist with 10+ years of experience in data science and a PhD in physics
A senior software engineer with 10+ years' experience (including MedTech)
A lawyer who helped scale one of Europe's largest hospitality platforms
A creative director with more than two decades in design & experience systems

We are backed by renowned investors, including Norrsken Evolve , one of the world's leading impact investors, and have an extensive network of healthcare leaders, founders, and key decision-makers behind us. Our first customers include some of Europe's largest hospital groups and high-performance clinics.

This is a role for someone who wants to build a product with real users, real constraints, and real impact—shaped by direct exposure to clinicians and continuous customer development.

What you will do

You will help build the ML backbone that turns messy, multimodal healthcare data into something reliable, comprehensible, and empowering for professionals who depend on it to treat patients safely. The focus: robustness, traceability, stability, and practical usefulness .

Core responsibilities

Architect and maintain the multimodal ML pipeline (speech, handwriting, structured forms, PDFs) powering our documentation co-pilot.
Design and refine hallucination-resistant extraction logic , including hybrid systems (models + programmatic constraints + validation loops).
Implement traceability features so clinicians always understand what the model inferred, why, and from where.
Build feedback loops that allow us to continuously learn from real clinical use—while keeping reliability front and centre.
Enable customisation for clinical teams (20% flexible surface) while keeping 80% standardised for safety and scalability.
Improve latency, efficiency, and throughput , not for vanity performance but to expand the breadth of workflows the system can support.
Contribute to interoperability pipelines with third-party systems (laboratories, hospital software, and service providers).
Stay current with LLM/ML research (agentic patterns, retrieval strategies, new architectures) and translate promising approaches into production features.
Work closely with physicians, nurses, therapists, and coding experts to understand failure modes and continuously improve output quality.
Shape the internal engineering culture: clarity, curiosity, ownership, and depth of craft.

What success looks like after 12 months

Your impact will be felt not in academic papers but in the daily life of clinical teams.

By the end of your first year, success means:

1. A robust, reliable ML engine powering real clinical workflows

Multimodal inputs are handled consistently and predictably.
Model behaviour is highly traceable, with clear provenance of every output.
Hallucinations are reduced to an operationally negligible level.

2. A system that knows when AI is the right tool — and when it isn't

Graceful fallback paths exist for uncertain outputs.
Confidence scoring and escalation logic are mature and trusted by clinicians.

3. The tool empowers healthcare workers

Doctors, nurses, and therapists can finish documentation faster without losing control .
Outputs feel precise, intelligible, and trustworthy.
Customisation feels natural without creating fragmentation.

4. Strong, continuous improvement loops

Clinicians give feedback because it's useful and visible in the product.
You've built automated evaluation mechanisms grounded in real clinical data and validated examples.

5. Interoperability is no longer a "future feature"—it works

Laboratories, external tooling, and healthcare software connect cleanly into our outputs.
Aiomics becomes an easy partner in the larger ecosystem.

6. Engineering foundations support scale

CI/CD for models is stable.
Monitoring and evaluation frameworks are reliable.
Latency is optimised enough to unlock new product surfaces.

This is the work that will make Aiomics the most trusted documentation intelligence layer in European healthcare.

What we're looking for

Essential

5+ years building ML systems in real production environments with real users.
Track record at a high-bar engineering environment (FAANG, top-tier startup, or similar) — we're looking for someone who's seen what "good" looks like at scale.
Strong experience in:MLOps & model lifecycle management Hybrid inference systems (LLMs + deterministic logic) Reducing hallucinations & implementing traceability Speech/OCR/NLP pipelines or equivalent multimodal systems Python
Comfort operating in ambiguity with high autonomy.
A product mindset: you think in terms of usefulness, not demos.
Curiosity for healthcare, willingness to learn from clinicians.

Our stack

Python, FastAPI, LlamaIndex, AWS (ECS Fargate). You don't need to know all of these — but you should be able to pick them up fast.

Nice to have

Experience with terminology-heavy domains (ICD, SNOMED, OPS, FHIR, openEHR).
Experience with complex extraction problems (e.g., logs, messy text, mixed modalities).
Prior work in regulated or safety-sensitive environments (healthcare, aviation, finance).
JavaScript/TypeScript experience.
German language skills.

Why join Aiomics

This is not a typical "AI startup" job. Here's what makes it different:

You work with real users from day one.

We are already integrated into some of Europe's biggest hospitals and advanced specialty clinics. You will work directly with medical directors, nurses, and therapists.

You join a genuinely senior team.

We're a small group of experienced professionals who value substance over hype. We work with discipline and focus — no theatre, no reinventing solved problems for the sake of it.

You get high autonomy in a small, powerful team.

We intend to stay lean. That means:

You shape architecture, roadmap, and process.
You build something enduring, not a prototype farm.
You can leave your signature on a product that changes everyday clinical life.

Your work improves lives directly.

Not metaphorically. Doctors go home earlier. Nurses have more time to actually provide care. Patients receive better-coordinated care. Health systems waste less time, money, and human energy.

You gain exposure to a serious network.

Through our investors and advisory circle, you'll meet some of the most influential people in European health innovation, policy, and tech.

You get to solve intellectually rich problems.

Where machine learning meets medicine, human behaviour, real-world data, and legal constraints. If you want an engineering challenge that matters—and to work with people who take the mission seriously, this is it.