Writing

Longer-form work.

Occasional pieces we've written up properly. Case studies, reports, technical walkthroughs. Not many, because writing something worth keeping takes a while. More may arrive when the subject warrants it.

Case Study August · 2025 35 pp ~35 min read

Project
Azgzaw.

Rebuilding an enterprise data platform on Microsoft Fabric. What worked, what didn't, and the lakehouse pattern we eventually settled on.

A full technical and editorial walkthrough of an eighteen-month engagement. Covers the ingestion layer from SAP Cloud and Infor Syteline, the medallion architecture we shaped around the business rather than the textbook, the configurable data quality engine we built from scratch, and the DirectLake semantic model underneath it all.

Written for practitioners. Assumes you know what a Power BI dataset is, but explains every Fabric-specific decision as we go.

Report July · 2023 67 pp ~60 min read

A Compendium
of MOT Tests
in 2022.

Every MOT test carried out in the UK in 2022, pulled from DVSA records. Pass rates, failure modes, what mileage really predicts, and the patterns nobody talks about.

We pulled every MOT test record the DVSA publishes, 41.6 million of them, and spent a few months trying to answer the questions we'd always wondered about: which makes fail most often, does mileage actually predict failure, which test centres are outliers and why, and how the picture shifts by region.

The goal was partly curiosity and partly to demonstrate the kind of analysis a capable Power BI and SQL engagement can produce when given time, clean data, and a proper question.