The problem
A regional healthcare operator was running three separate data systems: a legacy on-premises data warehouse, a cloud EHR integration layer, and an analytics database built by a previous vendor. The three systems diverged weekly. The analytics team spent 16 hours every week manually reconciling numbers before any report could be trusted.
TODO: copy from Zane — real pull quote from the client.
Our approach
We designed a cloud-native data platform on Azure that ingested from all three legacy sources, transformed data through a well-tested dbt layer, and served a single queryable surface to the analytics team. Compliance was a first-class requirement throughout.
Phase 1 — Inventory and data mapping
We catalogued all data flows, documented the divergence points, and mapped HIPAA-relevant data fields across all three systems. The client had never had a single document describing where their PHI lived — that alone was a compliance win.
Phase 2 — Ingest and transform
We built an Azure Data Factory ingest layer with encryption-at-rest, managed identities, and audit logs on every pipeline run. The dbt transformation layer gave the analytics team a testable, version-controlled model of their business logic for the first time.
Phase 3 — Analytics layer and handoff
We provisioned Azure Synapse Analytics and ran working sessions with the client's analysts, transferring ownership of the dbt models. We did not leave until the team could add a new data source without us.
Outcomes
The analytics team now has one source of truth. Reports that took two days to compile produce in under ten minutes. The platform passed the client's internal HIPAA compliance review on the first attempt.