AI Engineer

Health is the most precious thing we have.

At Palm, we’re building the founding team that will make it better for everyone, starting where it matters most: the first 1000 days of life.

📍 Rome, hybrid (2 days a week in the office)

💼 Full-time, mid-senior · 5+ years

💶 €40,000 – €60,000 gross per year (RAL), based on experience

THE ROLE

You’ll make Palm’s AI reliable, measurable and ready for a world of agents. You’ll take AI features from experiment to production with real autonomy, and you’ll own our evaluation practice: how we measure, improve and prove the quality of every answer. Working with our AI and Product Lead, our CTO, the software team and our clinical team, you’ll turn our technical direction into solutions that are testable and maintainable.

FOUR HARD PROBLEMS. YOU’LL OWN THEM.

01 · Messy clinical data. Letters, reports, photos. Turn them into structured, trustworthy data, and never lose the link to the original source.

02 · Evals are the product. Build clinical benchmarks with doctors: real questions, critical cases, red flags. Quality is measured on every release, not assumed.

03 · Built for agents. Design Palm’s capabilities so they can be used by our own products and by other AI agents, with cited sources and consent respected.

04 · A safety line you can test. Palm informs and orients, it doesn’t diagnose. Make that boundary measurable and automated, so that when something drifts we see it early.

WHAT YOU’LL DO

• Build and evolve RAG pipelines: ingestion and structuring, chunking, indexing, retrieval and reranking where it pays off.
• Improve LLM-based chat and workflows, including tool use and integration with application data and services.
• Ground every answer in approved clinical sources and the user’s relevant data, and check that the sources really support what the system says.
• Combine generative models with rules, structured validation and deterministic checks. Handle missing data, conflicts and uncertainty with abstention or human review.
• Build evaluation datasets and pipelines for retrieval, answers, document extraction and tool use, including critical cases and regression tests.
• Improve data extraction from documents, keep the link to the original sources, and make uncertain information explicit.
• Build automated pipelines for knowledge-base updates, with versioning, source traceability, grounding checks and monitoring.
• Work with our doctors to define approved sources, clinical requirements and acceptance criteria, and validate how the system behaves after every update.
• Find where errors come from (data, retrieval, model or application logic) and track quality, latency and cost.
• Write tested, documented code, review PRs, and help release and monitor in production.

ABOUT YOU

• 5+ years in AI/ML and/or software engineering, including shipping LLM and RAG solutions to production. We also consider candidates with fewer years if we think they could be a solid fit.
• Strong Python and solid engineering habits: Git, tests, debugging, APIs, backend integration.
• You know LLMs, embeddings, vector search and RAG well, and you can explain why you chose what you chose.
• Hands-on with agents and tool use.
• You design reproducible evaluations, analyse errors and show improvement against a baseline. Checking a few examples by hand isn’t evaluation to you.
• You understand hallucinations, prompt injection, output variance, source quality and uncertainty, and you care about data protection.
• You work autonomously, collaborate well and flag risks early.
• You have the right to work in Italy.
• You’re fluent in both Italian and English.

Bonus points: healthcare, sensitive data or regulated software; OCR and document understanding; benchmarks built with domain experts; GCP, containers, CI/CD, AI monitoring, orchestration frameworks such as LangGraph; knowledge graphs, rule engines or verifiable reasoning; a degree in CS, engineering, maths, physics or similar, ideally with a master’s or PhD in AI/ML.

YOUR FIRST PRIORITIES

Get to know our existing pipelines, set a quality baseline and tackle the main errors with measurable improvements. Help strengthen our evaluations and output checks, and build a sustainable process for updating the knowledge base.

LONGEVITY STARTS IN THE WOMB

The first 1000 days, from conception to a child’s second birthday, are the most formative window in human health. This is when the brain, the immune system and the metabolism are wired. Research on the Developmental Origins of Health and Disease (DOHaD) shows that what happens in this window shapes neurodevelopment, immunity and metabolic health, with effects that last a lifetime.

In the first 1000 days you can programme the system. In adulthood, you can mostly repair it.

WHAT WE’RE BUILDING

Health data is the most important data people own, and the worst organised. General-purpose AI can talk about medicine, but it can’t reason over your record or show where an answer came from.

Palm is the clinical memory of every person, with an AI on top that answers only from validated clinical evidence and the person’s own data. We start with children and the first 1000 days. The goal is a lifetime.

• Science first. Built on validated clinical protocols and guidelines, reviewed by doctors, never the open internet.
• Measured, not guessed. Every answer is linked to its sources and evaluated against benchmarks built with clinicians.
• Built for what’s next. Soon people will ask their own AI agents about their health. We want Palm to be the trusted source behind those answers.

WE’RE LOOKING FOR PALMERS

Everyone has a handful of names on speed dial: the family doctor, the partner, the colleague who knows all the tricks, the friend who picks up at 3am. We’re looking for those people: the ones others call first when something really matters. We call them Palmers.

We’re building our founding team, and we’ll be very selective: the first people shape everything that follows. When the question is AI, we’d like this candidate to be our first call.

WHO YOU’LL WORK CLOSELY WITH

• Our CTO and the talented Palmers of our tech team.
• Our CEO, a tech innovation leader with 15+ years in Big Tech (Yahoo, LinkedIn, Amazon, AWS). Part of the launch teams that brought Amazon to Australia and Singapore; at AWS she built a global innovation program for 500+ companies. As an Amazon Bar Raiser she has run 400+ interviews.
• Our CMO, a paediatrician and neonatologist (MD, PhD) with 25 years in medicine, from neonatal intensive care to research at Massachusetts General Hospital and Harvard Medical School, where she is still an affiliated researcher. She has published on AI in neonatal care.
• The paediatricians, neonatologists and health experts we partner with, and the Palmers in marketing, growth and business.

YOU’LL LOVE IT HERE IF…

✓ You’d rather prove a model works than demo it.

✓ You want to own a problem end to end, not a ticket.

✓ You think health products deserve a scientific standard.

PROBABLY NOT FOR YOU IF…

✗ “It seems to work” is good enough for you.

✗ You want a big-company roadmap handed to you.

✗ You are looking for remote-only opportunities.

HOW WE HIRE

1. Intro call (30 min): who you are, what you’ve built, what you’re looking for.
2. Technical deep dive with our CTO.
3. A real Palm problem. We work through it together. No whiteboard puzzles.
4. Meet the founders.
5. We assess technical preparation through education, professional experience and the projects you’ve delivered.

Believe you’re a Palmer?

👉 Click Apply and leave us your details. We’d love to know who you are, what you’ve built, and what makes you the person on everyone’s speed dial.