About
I design and build data and AI systems, with the goal of solving real business problems. That means I start with what we really want to achieve, and only then look for a technical solution, connecting business and IT from the data and AI angle.
What I work on #
I work at Schuberg Philis, where I started our current data and AI practice and where I both design the models and build the systems that run on them. Most of that work sits in the knowledge layer: RDF ontologies, knowledge graphs and semantic models that encode what business terms mean. On this blog I write about getting enterprise data ready for AI and putting it to work.
Before that I spent about a decade on the data underneath. For an industrial IoT platform covering 98 plants, 4,000+ machines and 90 billion data points, I designed the ISA-88 and ISA-95 models and built the real-time event processing in Go. I led an Azure data platform in a regulated financial environment, with automated lineage and auditability. Along the way I modeled GraphQL schemas to connect a parcel logistics chain, built Data Vault 2.0 architectures for national KPI reporting, and put a data quality framework in place together with the data stewards who had to work with it.
AI has already changed how I work day to day, and connecting models to business data has convinced me the same is coming for how businesses operate. The bottleneck is not capability but trust. Trust means being able to check an answer against the data it came from, which requires something explicit to check it against. Constraining a generative model with deterministic domain logic, the neuro-symbolic approach, is what makes an answer checkable. A knowledge layer is where that logic lives. It only works when the data underneath is governed, the lineage is traceable and the quality is known. This is why I work on the knowledge layer and the data underneath it.
Background #
- MSc Information Studies, University of Amsterdam
- BSc Bedrijfskunde (Business Administration), University of Groningen
- Based in Amsterdam
How organizations work interests me as much as the technology does. Modeling an organization’s data is a direct way to see what it is really trying to do, and a practical place to start changing it.
If any of what I write is useful (or wrong), I’d like to hear it. My links are below.
Everything here is my own opinion and does not represent Schuberg Philis.