Are you an engineer who's seen enough agent demos and wants to put one to work?

Crunch builds the data-driven companies of tomorrow. And tomorrow, a fair chunk of the analytical work in those companies gets done by agents: answering questions, monitoring operations, flagging problems before a human would have spotted them. We are looking for the person who builds those agents. The ones that survive contact with real data, real users, and real Mondays.

That means engineering discipline first, and hype tolerance of exactly zero.

As an Applied AI / Agentic Engineer, you build AI systems on top of the data platforms we run for our clients. Think conversational BI that doesn't just answer questions but knows which questions to keep asking. Think agents that watch a supply chain across sales, planning, and operations, and speak up when something drifts. You know that a good agent is 20% model and 80% engineering: tools, context, evaluation, guardrails. You also know when the honest answer is "this shouldn't be an agent, it should be a query."

In your role, you work at the sharpest edge of what we sell, which means what you build one month becomes what we demo the next.

What will be part of your responsibilities?

• Design and build LLM-powered applications and agents on client data: conversational BI, monitoring agents, decision-support tools;

• Connect agents to the data layer properly: semantic layers, tool interfaces, structured context. No prompt-and-pray;

• Build evaluation and monitoring so we know an agent works, instead of hoping it does;

• Turn prototypes into production systems together with our data engineers;

• Keep up with a field that changes monthly, separate signal from noise, and share what actually matters with the team;

• Demo your work to clients and prospects. If it impresses a skeptical supply chain director, it's good;

• Help define what the agent layer of our platform becomes. This role has more blank canvas than most.

Our offer

First off, you will become part of a steadily growing team of people keen on applying the latest technologies and techniques in the fields of data science, machine learning & AI. You will work in a vibrant, innovation-loving environment with the proper dash of geekiness and love for board games.

While you will be able to work both at the office (in Gent or in Rotterdam) or at home, your presence might also be required on-site with clients. Our offices in Ghent and Rotterdam are easy to reach, located in a colorful neighborhood of each city center, close to a vast array of after-work shenanigans.

We provide an attractive salary package stuffed with additional benefits such as an (optional) company car & fuel card, smartphone, and other gear. We help you to get around either by car, train, and/or e-bike. It's yours to choose.

We put a lot of time and effort into your personal growth and development, both as a technician and business professional. We provide dedicated feedback and coaching sessions to help you walk the career path you envisage. Expect your own talent development roadmap and the opportunity to showcase your knowledge to a broader engineering community.

On top of that, we do team trips, planned events, ad-hoc events, anything really to ensure you shine within a thriving team.

Up for the challenge?

Apply now and we'll review your application, and may invite you for an initial phone interview. A job-specific challenge and presentation later, you could be the one who makes "AI agent" mean something around here!

Required Skills

• Strong software engineering fundamentals in Python; you write code others can maintain;

• Hands-on experience building LLM applications beyond a wrapper around an API: tool use, retrieval, structured output, evals;

• A pragmatic view on where agents add value and where they're theater;

• The ability to explain to a client what the system can and cannot do, honestly;

• You are fluent in English.

Preferred Skills

• Experience with MCP, agent frameworks, or building your own orchestration (bonus points for opinions on when frameworks hurt);

• Familiarity with semantic layers, RAG architectures, or conversational BI;

• Experience with Databricks or modern data platforms;

• Enough front-end skill to make a prototype people want to click on;

• Speaking Dutch is a definite plus. French is also nice to have.

Above all, you'd rather ship one agent that quietly does a job every day than ten demos that get applause and gather dust.