Building the Data Foundation for a University in Motion

"We partnered with Cegeka to make a fundamental shift," says Jeroen Bovend'eerdt, Product Owner BI, Analytics & Data Coordination at Eindhoven University of Technology (TU/e). "We moved from a traditional BI approach where every data request filtered through my team, to one that treats data as a strategic asset, with Databricks as the foundation and self-service at its core."

Few universities in Europe are as deeply woven into the fabric of industry as TU/e. Founded in 1957 in what was then Philips' backyard, the university has spent nearly seven decades producing the researchers who power some of the continent’s most advanced companies. ASML, DAF, Vanderlande, Philips and thousands of suppliers, combined in the Brainport ecosystem, are still on its doorstep today. With 13,500 students, 5,000 employees and nine departments, TU/e holds itself to high standards.

That standard now carries higher stakes than ever. TU/e plays a pivotal role in the Beethoven programme — a €2.5 billion Dutch national investment to secure Europe's position in the global semiconductor industry. To feed the talent pipeline that companies like ASML urgently need, the university must expand its programmes and research facilities at pace, and to steer that growth with confidence, it needed to trade legacy reporting for a modern data strategy.

Moving Beyond the Queue

Since 2018, TU/e had relied on a centralised BI programme built on a fairly traditional Microsoft SQL data warehouse, supporting departments such as HR, finance, research and facilities. But the central BI team found itself perpetually overwhelmed.

"The BI team found itself perpetually overwhelmed. Demand always exceeded capacity."

"Demand always exceeded capacity," Bovend'eerdt recalls. "We were constantly prioritising and playing catch-up. Every report, every dashboard, every data request joined the same long queue. It was a problem of architecture as much as capacity — adding headcount alone was never going to solve it."

Strict privacy compliance requirements added further pressure to find a structural solution.

The Self-Service Shift

The answer wasn't incremental; it required a fundamental rethink. Working with Cegeka, TU/e built a new platform with Databricks on Azure, and pivoted toward a data mesh model in which departments build their own data products.

"Everything had to be built directly in Databricks: one coherent platform, minimal tooling diversity. Cegeka agreed, and delivered."

"In practical terms, this means true self-service," Bovend'eerdt explains. "Any department with Power BI knowledge no longer needs to submit a request and wait in line. They build and publish their own reports. They are now in the driver’s seat, determining their own development speed. Our role has shifted from bottleneck to backbone: we unlock the data, set the standards, and maintain the infrastructure rather than serving reports one by one."

The platform was built for the long horizon, with sandbox environments for researchers and genuine AI integration from the start. "These are not nice-to-haves for an institution that takes data science seriously," Bovend'eerdt says.

Combing through CVs

The choice of Cegeka as a partner reflected the university’s own technical culture. "We really combed through those CVs and spoke at length with the technical specialists," Bovend'eerdt explains. "Cegeka’s profiles were impressive and helped seal the deal.”

"Cegeka’s profiles were impressive and helped seal the deal."

What also sealed the partnership was Cegeka’s willingness to work within the TU/e vision. Where a standard approach would have included Azure Data Factory for data ingestion, TU/e insisted on a single, coherent platform built directly in Databricks. “We wanted everything in one place with minimal tooling diversity. Cegeka agreed and delivered,” says Bovend’eerdt.

The resulting migration was a complete re-architecture: dismantling a legacy environment built on SQL Server, Visual Studio, and Azure Data Factory, and rebuilding it as a full Databricks data platform. Throughout the process, everyone worked shoulder to shoulder. "Extremely knowledgeable, flexible, and frankly reachable at all hours," Bovend'eerdt says of the Cegeka team.

Faster, Cleaner, Compliant

The outcomes are already visible. Departments now set their own pace, and data quality has improved because ownership creates accountability. Teams can also share datasets directly with one another without beeing heavily dependant on the central team.

The system is now fully privacy-compliant. “That compliance is enabled by an AI-powered classification tool embedded in the Databricks platform," Bovend'eerdt says. "It automatically scans datasets for privacy-sensitive information, not just in structured fields, but in unstructured ones too, including free-text comment fields. Where sensitive data is found, it is pseudonymised, with access configurable by role."

With the platform in place, TU/e is already building a dedicated Beethoven dashboard to track its role in the semiconductor talent pipeline. "It gives our Board a clear view of where we stand against specific Beethoven goals — budgeting, student retention, and so on. This is significantly easier and faster than in the old setup.

We are now ready to put AI Agents at work to build selfservice datamodels at lightning speed and help end-users easily find the reports that they didn’t know existed!