A reporting model rebuilt on Fabric for a university division, then coached into the team

A data model that had grown hard to change, several sources of different shapes, and a team that knew Fabric and wanted a pattern they could repeat themselves.

One representative slice of the data built end to end: ingestion into raw, transformed and curated layers, a star schema, incremental loading, monitoring and data-quality checks, and a Direct Lake semantic model, all under one naming convention, so the team could read the shape of it and repeat it.

The team's analyst was coached through it notebook by notebook, then built the remaining sources the same way.

A reference implementation, documentation the team can open in a browser, and the remaining sources built by the team itself.