CASE STUDY · data engineering · 2026

Strata

Public medallion data pipeline on synthetic insurance data — DuckDB, dbt, SCD2, data contracts, 42 tests, CI green.

uilt as the public proof of NDA-locked DWH work at BRILife. Full medallion architecture (bronze→silver→gold) on synthetic insurance data. Implements SCD2 slowly changing dimensions, data contracts, reconciliation, and 42 automated tests — all patterns used in production but open-sourceable here.

highlights

+Medallion architecture: bronze/silver/gold layers with clear promotion rules
+SCD2 implementation for policyholder dimension changes
+Data contracts enforced at ingestion boundary
+42 tests covering transformation logic and reconciliation
+CI pipeline (GitHub Actions) — green on every commit

what was hard

Designing synthetic data that mirrors real insurance domain without exposing BRILife patterns; Making SCD2 testable in a local DuckDB environment without a full warehouse