Senior Solution Architect — Data & Integration
I design the data and integration backbone, build what runs on it — and get the organisation to adopt both.
20+ years architecting the data, integration and streaming platforms that enterprises run on. Almost none of it came with a mandate behind it: 18–20 product teams onto one platform, three estates onto one event contract, a finance function that had to trust a number before it would use it. Not slideware — systems in production, and the decisions written down where people could argue with them.
Track record
Four numbers from work that shipped. Each one links to the case study behind it, including the part that was argued about.
18–20 teams, no mandate
Product teams consolidated onto one API platform by making the paved road cheaper than staying put — ~€250–300k/yr saved
3 estates, 1 contract
SAP, legacy ESB and cloud teams brought onto one event contract — agreed before either side wrote code
~30+ company codes
SAP Finance ledgers into Snowflake — adopted because Finance could reconcile the numbers itself
~500M+ req/month
Federated cross-cloud API platform across AWS and Azure (Cloud Gateway)
How I get it adopted
The design is usually the easier half. What decides whether anything ships is the teams who each already have something that works, the function that has to trust a number, and the contract somebody signed four years ago. Three situations, with the disagreement and what it cost.
Twenty teams, no authority to move any of them
Asking for a mandate would have produced twenty exceptions. The platform had to be cheaper for the team instead, and the first teams to move were the ones with the worst existing setup, not the easiest to convince.
Three teams, three definitions of “done”
For SAP the job ended at “the events are on the broker”; for the cloud team it started at “we consume what is there”. Everything that matters sat in the gap. I wrote the seam down and took the boring half myself.
A finance organisation with no reason to believe me
Explaining the pipeline changed nothing. A reconciliation Finance could run themselves against their own ledger changed everything. Adoption followed the check, not the presentation.
What I build
Five areas, ordered by what I am usually engaged for. Data and integration lead. AI is last, and that is deliberate — a profile that opens with AI usually means the rest of the estate has not been thought about.
Data & lakehouse
Medallion lakehouses, contracts at the source seam, and CDC pipelines that stay reliable — built so the numbers reconcile against the system people already believe.
Integration architecture
The backbone that lets enterprise systems talk: legacy ESB estates decommissioned wave by wave onto event-driven, API-led, domain-owned platforms.
Event-driven & streaming
Kafka and broker-based integration as the spine of the estate, with schema evolution treated as a governance contract rather than a serialization detail.
APIs & gateways
API platforms that scale across dozens of teams — gateway strategy, one security model, and the developer experience that is what actually gets them adopted.
AI & automation
AI integrated the way any other supplier system is: behind a contract, with an evaluation gate before release, and the model kept outside the runtime so it stays replaceable.
See the code behind it
Public repositories you can clone and run — the “active customer” question, the seam decision, the Fabric-or-Databricks decision, the modernization lab, the streaming platform and the identity service.