Case study · August 3, 2026 · 3 min read
A Hospital Analytics Platform Built to Survive a System Failure
Building a reproducible, governed data warehouse that can be rebuilt from scratch in one hour

The client
Before this engagement, a Toronto-based specialty hospital serving 3,000 patients had no consolidated data platform. Reporting was pulled manually from six or seven separate clinical and operational systems. There was no single source of truth, meaning any executive question required days of reconciliation across siloed databases.
The hospital’s Head of Data and Strategy knew they needed a reliable foundation. The goal was to give leadership timely, trusted information for resource allocation and performance measurement. Momo Analytics was brought in to engineer the data warehouse that would make this operational.
The challenge
This was a partnership of organizational strategy and technical execution. The hospital defined the priorities; Momo Analytics built the production platform.
Here is how we engineered it for absolute resilience:
This was a partnership of organizational strategy and technical execution. The hospital defined the priorities; Momo Analytics built the production platform. And we did it with a single engineer.
-
Centralized, automated ingestion: Instead of relying on a fragile web of manual extracts, we consolidated data from all six source systems into a single pipeline codebase. We wrote dedicated Python ingestion jobs for every system—whether pulling from a relational database or an EMR extract—and landed everything safely into a unified staging layer.
-
Transformations and Infrastructure as Code (IaC): We used dbt to handle the data transformation layer. Absolutely nothing was done by hand. Every pipeline, every data model, and every business rule was defined in code. Every change was forced through version control (Git) and a strict CI/CD pipeline before it ever touched production data. This guaranteed that the entire warehouse was fully auditable and reproducible.
-
The one-hour, clean-slate rebuild: Because the ingestion logic, data transformations, and analytical layers all lived in version-controlled configuration, the platform didn’t rely on anyone’s memory of how it was set up. If the underlying infrastructure melted down, we could spin the entire environment—source connections included—back up from a clean slate in about an hour. We didn’t just write a disaster recovery document; we proved it by tearing the environment down and rebuilding it from scratch with zero data loss.
A- ccessible reporting: We wired the clean, governed data up to Tableau to power the hospital’s annual board balanced scorecard and operational dashboards. The reporting layer simply sits on top of a rock-solid foundation, using tools the hospital can operate without needing to retain a specialist indefinitely.
The outcome
The hospital moved from fragmented, fragile reporting to a governed data warehouse that practically runs itself. It feeds the annual board scorecard, handles daily operational reporting, and most importantly, has a proven, one-hour disaster recovery time.
Eventually, this exact same robust architecture became the technical foundation that the hospital’s massive Epic EMR migration ran on top of.
They didn’t just get a set of dashboards. They gained resilient, reproducible infrastructure built on strict engineering standards, designed to be safely maintained and instantly recovered by the people responsible for it.
Read the next step in the evolution: Migrating a Legacy EMR to Epic With Zero Data Loss.