AI Engineer

We're hiring an AI Engineer who will grow into owning all AI-generated content on the Lumoo platform. You'll start with clear priorities and close support from our CTO and Head of AI, and within three to six months we expect you to take broad, independent responsibility for the quality and direction of everything our models and Iris produce.

About Lumoo

Lumoo already runs AI image and video generation in production for some of Europe's largest fashion, sport and retail brands. Our platform covers enterprise content production, a self-serve tier for e-commerce merchants, and a virtual try-on SDK. Our architecture lets brands create on-brand content across their whole organisation, with brand environments that keep every generation consistent with each customer's visual identity. On top of that we are developing Iris, our agentic AI that increasingly orchestrates the production flow.

Your first months

You'll work across the full production chain, from our generation architecture and brand environments to Iris and the images and video a brand publishes. Most of our stack is built on pre-trained foundation models, so the hard problems sit in how we run them: how prompts, system instructions and multi-step workflows are versioned and rolled out, how Iris and our pipelines pull in the right brand data and assets through retrieval, and how we route traffic between models and providers.

A big part of the early work is observability and evaluation. Our models never give the same answer twice, and you'll help us track that behaviour in production, catch off-brand or failed output before a person sees it, and follow generation costs, latency and safety guardrails. We expect this to mean building our own monitoring system, and you'll be central to designing it.

Where the role is going

Over time you'll own the question of how we operate AI output for the best possible quality and performance, continuously and as models, customers and volumes change. You'll also become the person we turn to when a customer asks what Lumoo can do. That means knowing what our image, video and Iris capabilities deliver today, and being able to say what is realistic to build next and how long it would take.

Research is part of this from day one. The field moves fast, and you'll test new models, techniques and agent designs against our pipelines and decide what belongs in production. Expect the job description to change as the role grows. The work will get broader, more abstract and more self-directed, and we want someone who is excited by that.

Who we're looking for

This field is new enough that very few people have years of experience in it. We care more about how fast you learn and how much you want to own than about a long CV, and we think one to two years of relevant experience can be enough if you are ready to take on senior responsibility quickly.

• Experience automating operations around stochastic processes in production. This matters most to us. You've built monitoring, evaluation or recovery for systems where the same input can give a different output, and you know what it takes to make that dependable.
• You have built or fine-tuned your own models and understand how data, base model and training setup affect output quality.

You'll also need to be comfortable working without a detailed brief, setting your own priorities and explaining technical trade-offs to people who aren't engineers, including customers.

It's a plus if you have worked with prompt versioning or LLM evaluation tooling, vector databases and RAG, agentic systems, GPU inference serving, product or fashion imagery, or GDPR-sensitive data pipelines.