Software Company in Leuven

Leuven has a particular kind of software problem. The research is excellent, the prototype demonstrably works, and then a customer asks whether it can run every day without a PhD student restarting it. Between those two states sits ordinary, unglamorous engineering: authentication, tenancy, error handling, data retention, deployment, monitoring, and an interface someone can use without a walkthrough.

That gap is what Rui Codex builds across for spin-outs and research-driven businesses around KU Leuven and the imec ecosystem. We are not the people who invent your algorithm — you have that. We are the people who make it a system your first ten paying customers can depend on, without rewriting the part that already works.

What we typically build for Leuven spin-outs:

• Productisation — turning a working prototype into a deployable multi-tenant application with accounts, roles, and an operational story, on ASP.NET Core.
• Customer-facing platforms — dashboards and portals that make the underlying research legible to a non-expert buyer, which is usually what closes the deal.
• Integration and data pipelines — getting real customer data in and results out, reliably, on a schedule, with failures that announce themselves.
• AI engineering — putting models into production properly: versioning, evaluation, cost control and graceful degradation when the model is wrong.

We work as an external engineering team with the founders, and we write the handover documentation on the assumption that you will hire in-house engineers later — because you should, and the work should not become a dependency on us.

Will you rewrite our prototype?

Not if we can avoid it. The research code usually contains hard-won knowledge that a rewrite quietly discards. The normal approach is to wrap it — put a proper service boundary around it, and rebuild only the parts that genuinely cannot scale.

We have grant funding with fixed deliverables. Can you work to that?

Yes, and it is common here. We scope to the milestone and are explicit about what is production-grade and what is deliberately a demonstrator, so the distinction is written down rather than assumed — which matters a great deal when the next funding stage reviews it.

What happens when we hire our own engineers?

That is the intended outcome. We document as we go, keep the stack conventional so it is hirable, and hand over. Several clients have gone from us building everything to us reviewing pull requests to us not being needed, which is what success looks like.

Can you sign an NDA and work with sensitive research data?

Yes. NDAs are routine, and where the data is genuinely sensitive we work with anonymised or synthetic datasets during development and touch production data only under agreed controls.

How small a project will you take?

Small enough to prove we are worth working with. A first engagement is often a few weeks on one well-defined piece — an integration, a dashboard, a deployment pipeline — rather than a platform commitment.