We build things that work
in production.

I studied chemistry. I didn't come through the conventional route — no computer science degree, no data engineering bootcamp. I came through project management and project accounting at Mott MacDonald, one of the UK's largest engineering consultancies, where I ran headfirst into a problem that turned out to define the next decade of my career.

Month-end reporting took two weeks. Not because the data was hard to find. Because nobody had thought carefully about how it was structured. The organisation was sitting on everything it needed to make fast, confident decisions — and producing those decisions fourteen days too late to act on them.

So I taught myself. Late nights with Power BI before most people in the company had heard of it. Kimball dimensional modelling. Power Query. The principles underneath the tools, not just the tools themselves. I built an Excel data model from scratch — and the two-week wait became hours after month-end close. Same data. Same team. Completely different outcome. That result opened doors that led directly to everything that followed.

What I understood then — and have applied ever since — is that the problem is almost never the data. It's the architecture. Bad structure creates lag, distrust, and manual workarounds that calcify into permanent process. Good architecture compounds. Today, that means Microsoft Fabric — the platform where the entire data estate, from ingestion to insight, lives in one place and can be engineered properly. Power BI and Fabric are where I do my best work. Not as a tool operator, but as someone who understands how the whole system should fit together.

AI sits alongside that work — not as a replacement for it, but in genuine symbiosis with it. The most interesting problems I encounter now live at the intersection: data platforms that are structured well enough for AI to act on, and AI systems that make data platforms more powerful in return. I build agents and automated workflows, but the thinking behind them is the same as it's always been — understand the constraint, remove it structurally, and build something that runs without hand-holding.

“DATAXYZ is a bet on a different way of doing this work — where the outsider instinct, the systems view, and the discipline to build things properly are the whole point.”