I led the implementation, governance and day-to-day operation of a Power BI / Fabric analytics platform built on SAP data — from first use to production, as the single owner of security, refresh reliability, adoption and documentation for roughly 50 users. The case studies below walk through the platform decisions behind that work: what was considered, what was rejected, and why.
Case Studies
Rejecting DirectQuery: choosing Import mode for the primary semantic model
A CDS-view DirectQuery approach looked appealing for freshness, but testing showed it was too slow and too expensive under real usage. Here's the decision record for moving to a secured Import-mode architecture instead.
Row-level security at scale: governing access for ~50 users with one workspace
How group-based RLS and a Fabric app structure kept report access aligned to role without a bespoke build per team.
Governance as documentation: naming, standards and the operating model
Working solo meant the "operating model" had to live in documentation — naming conventions, workspace structure, release process.
Monitoring platform health: capacity, refresh reliability and performance
Using usage metrics and performance analytics to catch degradation before users did, and to guide ongoing optimisation.
Report Craft
Sales Quantity report — full interactive rebuild
A working rebuild of a production Power BI report's UI: live filters, sortable and drillable tables, and interactive charts — built as a lightweight native web app, not an embedded Power BI report. Customer names and figures are fabricated.