I build reporting that holds up when someone checks.

I have worked with data for 15+ years, most of it in financial services, and most of that on the kind of reporting that has to satisfy two audiences at once: an executive who needs the answer, and an auditor who needs to know where the answer came from. Those two demands pull in opposite directions more often than anyone admits, and reconciling them is most of the job.

A lot of business intelligence work is judged on whether the dashboard looks good in the demo. I am more interested in what happens eighteen months later — when the person who built it has moved on, a number has moved and nobody can say why, and the model turns out to be a file on somebody’s laptop. Most of what I do is aimed at that moment rather than the demo.

In practice that means dimensional models with documented grain, measures that are tested rather than eyeballed, lineage you can follow from a figure on a page back to the row it came from, and access controls that hold. It also means writing things down. A report nobody can hand over is a liability with a nice colour scheme.

Client work sits behind NDAs, so rather than describe it I rebuild the same disciplines on public data and publish the whole thing — the ETL, the model, the measures and the mistakes. The portfolio is six of those you can open end to end, and the writing is what they taught me.