Problem
A US retailer needed fulfillment data in SQL Server, but the existing numbers did not match the legacy system.
What I built
ShipHero → SQL Server ingestion via n8n and GraphQL into staging tables, stored procedures building silver and fact tables, and a parity audit against the legacy production system.
Key engineering decisions
- Staging then stored procedures, so every transformation is inspectable in SQL.
- Explicit handling of voided shipments and multi-cause data inflation found during root-cause debugging.
- Column-by-column parity audit before cutover, not sampling.
Validation and QA
Parity verified column by column: 68,890 orders, 25 columns, alongside the client’s QC team.
Result
Nightly sync the finance team trusts. Delivered through TheProject19.