Body of expertise
Twenty-five years and 230+ published articles, organised into the domains I actually work in. Each one is a thread I’ve thought through in depth, in public. Structure, applied to my own expertise.
Deep dives
The domains where my own original ideas live — each its own page, with the diagrams and the articles behind it.
Parse SQL into metadata, generate it back — and prove a migration identical, CTE by CTE.
Certified, experienced — and critical. The principles without the dogma.
Business-first, bi-temporal, pragmatic — the shape chosen for the job.
The diagram is a view, not the model. Generate it, never draw it stale.
Model as code in your IDE — the loop PowerDesigner never closed.
Executable metadata: design once, generate the SQL, tests, docs and lineage.
My core domain: designing the layers, patterns and integration that make a platform reliable by construction, not by heroics. The town-plan beneath the buildings.
A deep specialism: treating metadata as the executable heart of the platform — explicit, governed, machine-readable. The thread running through almost everything I build.
My strongest conviction: good models start with the business, not the database — and modeling is as much about communication as notation. Business-friendly mapping and universal models that business, engineers and AI can all share. Shaped by Alec Sharp’s masterclass.
A deep specialism — especially bi-temporal modeling: capturing how data changes over time and when we knew it, correctly. Data Vault principles without the dogma, dual / bi-temporal SCD2, and deterministic historization — true history you can actually trust and query.
Migrating legacy SQL to modern platforms — and proving the result is identical. CTE-by-CTE regression testing, deterministic generation, and validation that catches what final-output checks miss. See the signature engagement →
Treating SQL as structured metadata, not text: parsing legacy SQL into logical building blocks, and generating clean, deterministic, multi-dialect SQL back out. Write the intent once, generate the code reliably.
Reliability as a first-class concern: data quality as metadata, functional (not just code-based) lineage, governed corrections, and validation built into the pipeline. The wedge that makes the rest trustworthy.
Where my newest work lives: making AI reliable for data work through structure and rules, not hype. Metadata copilots without the magic, AI-assisted modeling and testing, and the validation loop that keeps generated output honest.
One of the things I’m best known for: making the complex clear. Architecture, data models, processes and lineage turned into diagrams people actually understand — hand-crafted in yEd and draw.io, and (my real specialism) generated automatically from metadata so the picture never drifts from the truth.
Hands-on, deep expertise with enterprise data-modeling tools — and the experience of pushing them well beyond their defaults. At de Volksbank I customised SAP PowerDesigner to generate physical mappings and SQL from logical models, and to parse SQL back into the model for full lineage. I also work extensively with erwin Data Modeler.
The point isn’t the tool — it’s getting any tool to serve a model-driven, metadata-first way of working.
These are highlights from 230+ published articles. The full archive lives on Medium. Where to read →
That’s exactly the kind of conversation I’m here for.
Start a conversation