An architect who still builds

I started where most engineers start - writing code alone, learning by breaking things. Early in my career I gravitated toward distributed systems and eCommerce, working across Magento, Salesforce Commerce Cloud, and eventually the composable commerce ecosystem.

That conviction came from watching too many technically elegant systems fail because nobody asked whether the business could actually operate them. I learned early that the architecture isn't the diagram - it's the set of decisions that let a business move at the speed it needs to.

At LiveArea, I led a team that grew to over 50 engineers, delivering enterprise eCommerce programmes for brands including Pandora, La Prairie, Birkenstock, Mizuno, and Daily Mail. That's where I learned the hardest part of architecture: it's not designing the system, it's aligning 40 people around a shared understanding of why the system works this way.

As CTO at SpinUp Digital, I was responsible for a 45-person engineering organisation, delivering commerce platforms for brands including Le Creuset and Air Malta across a mix of enterprise and high-growth SMB clients. Every one of them reinforced the same lesson: the teams that shipped well weren't the ones with the most sophisticated architecture - they were the ones whose architecture was legible, change-tolerant, and honest about its trade-offs.

That period also marked a shift in the kind of work I took on. I moved beyond eCommerce into full-scale digital transformation - event-driven architectures, large microservice estates, and eventually AI agent integration. It's where my work on agentic AI moved from theory into practice, and it set the direction for everything I do now.

Today, everything I do is focused on one problem: helping enterprises take AI from strategy to production. Everything I've built over two decades - distributed systems, platform migrations, scaling engineering organisations - is the foundation that makes production AI possible. As a maintainer of the Model Context Protocol, I'm helping shape the standard that connects AI agents to real-world systems. I bring that knowledge into every engagement - not AI for its own sake, but AI that teams can trust, govern, and measure.

Along the way, I learned that the best AI delivery comes from teams built around the architecture itself - senior engineers with full context, not borrowed capacity. Several of the teams I've assembled are still with their clients years later.

That's the through-line. Whether I'm defining a migration roadmap or reviewing a pull request, I stay close enough to the work to know whether the architecture is actually serving the people who have to live with it.