— Build
AI delivery & architecture
Production AI systems with the architecture you can audit. Agent topologies, retrieval, MLOps, and the AI Act §9 risk file authored alongside — not after.
Twenty years leading IT through every shift, and keeping it pointed where the business needs to go.
A federal health insurer · the capability
At a Belgian federal health insurer I built it from scratch: the reference models, the viewpoints, the governance, and the first architects to run them. Then I secured the funding to keep it running as a permanent capability, and led the IT-strategy work that set its direction.
A federal customs administration · the programme
Multi-million-euro delivery across the European customs union. I mapped the dependencies between the member-state systems and ran a steering process that actually made decisions.
A social secretariat · the turnaround
I took over IT for a legacy payroll platform that was failing. We cut its infrastructure cost in half and quadrupled the software's performance, both measured. An agile transformation cleared about five years of backlog in two months.
Now · the AI era
Machine learning, computer vision, deep RAG, agentic AI. Knowledge assistants and Claude Code run in my delivery pipeline, and I work with the AI Act as something I have to apply day to day. I finished the Postgraduate in AI Technology Architecture in 2026.
Modern Business-Intelligent Architecture and Solutions. Leadership in the age of AI.
Two decades of senior IT leadership in the Belgian public sector. I also train TOGAF and Agile.
Principal architect · Belgium
Twenty years of lived enterprise architecture. Infrastructure, data, code, and AI. Where systems that don't behave the same way twice meet the governance they still have to pass.
Architecture capability built from zero at a Belgian federal health insurer. Programme and portfolio delivery across the European customs union. A national payroll platform pulled back to stable delivery. One senior IT leader, accountable.
Agent topology: a planner agent orchestrates three executor sub-agents (retrieve, reason, act) via task-delegation, with a shared tool layer (DB, VEC, LLM, API) accessible across executors and a post-market monitoring overlay (log node) per AI Act Article 12. The diagram emphasises planner-executor orchestration with task-completion feedback.
— Build
Production AI systems with the architecture you can audit. Agent topologies, retrieval, MLOps, and the AI Act §9 risk file authored alongside — not after.
— Advise
EA-practice Kickstart, Author, or Nurture. Programme rescue. Ad interim senior IT leadership — I lead, reorganise, and manage out of crisis, then prepare for succession.
— Govern
AI Act, GDPR, COBIT, ISO 27k governance. TOGAF and SAFe training. Architecture decisions built to pass CISO and DPO review.
One senior IT leader, accountable. Build engagements run end-to-end; Advise engagements work in Kickstart, Author, or Nurture shapes. When an ad interim mandate is in flight, new Build engagements and Training cohorts pause or reduce. Book a conversation to scope.
A short scoping conversation is the cheapest way to find out whether MOBIAS is a fit. Thirty minutes, no slide deck, two architects talking shop. Pick a slot that suits you.