A fast-growing European software group serving tens of thousands of trades and construction businesses across Germany, Austria and Switzerland is building a single governed data layer underneath its existing products: one shared model that data from every product conforms to, serving both reporting and AI. This is one of the data team's first hires, built doers first with the architecture still being decided directly with the CTO.

You'll own that model. Data lands raw from every product, you decide what it becomes: defining what a customer, an order or a job actually is, one coherent answer across products that each have their own defensible version of the truth, and getting their owners to agree with you. The hard part isn't the code, it's holding that decision under delivery pressure and designing something the next product can absorb without a rebuild.

What you'll be doing

• Define the canonical model for the entities the business actually runs on, and map each source system into it.
• Reconcile the model back against its source and be the person who catches where conformance quietly loses information, before anyone else does.
• Work directly with domain experts to elicit and document the real semantics of their data, and get agreement from people who don't report to you and don't always agree with each other.
• Treat the model as software: tested, version-controlled, rerunnable transformations, not one-off notebooks.
• Design across a layered architecture, from raw landing through a conformed layer to serving.
• Set a technical direction that becomes everyone else's constraint, and be ready to document and defend it.
• Use agentic coding tools on real, complex work: decompose it, iterate, and review what comes out against the same bar as any other code.

What we're looking for

• 6+ years in data engineering, including having built a canonical or conformed model across genuinely heterogeneous sources in production, and lived with it afterwards, not a single warehouse over a single application.
• Deep dimensional and domain modelling experience, and the ability to talk through a modelling call that turned out wrong and what it cost to fix.
• Real reconciliation experience: taking a derived model back to source and being held to the result.
• Strong SQL and strong transformation engineering.
• Comfortable on a cloud platform without needing to own it.
• Genuinely comfortable working with domain experts rather than around them, this is as much a negotiation and communication skill as a technical one.
• A high bar on data protection, tenant separation, governance and lineage.
• Hands-on, established use of agentic coding tools on complex work.

Nice to have

• Experience building serving layers for both reporting and AI retrieval over the same model.
• Data contracts, schema evolution or versioning a model that other teams already depend on.
• Having modelled a domain through an acquisition or consolidation, under real deadline pressure.
• Experience in trades, construction or a comparable vertical.
• German fluency.