Cloud Data Platform Migration

The client ran their enterprise data warehouse on an on-premises SQL Server that had served them well for years, but the fixed hardware was reaching its limit. Years of changes had grown the ETL into a sprawl of interdependent, untested stored procedures with little structure, so every new requirement meant more technical debt and less focus on the value. With everything running on a single fixed server there was no way to spread the load, making complex job dependencies hard to orchestrate. The platform’s age also capped what the analytics and data science teams could realistically build on top of it.

I led the migration of the warehouse from on-premises SQL Server to a cloud-native stack on AWS: Snowflake as the data platform, Airflow for orchestration and dbt for transformations. I re-engineered the legacy stored-procedure ETL into modular, version-controlled and tested dbt models, built the AWS ingestion and Airflow orchestration feeding Snowflake, and planned and ran the cutover so the business kept reporting throughout the transition.

The result was a warehouse that scaled with demand. Reporting that used to take tens of minutes now returns in under a minute, new data products ship in days rather than weeks, and several analytics and advanced-analytics use cases were unblocked. Decommissioning the on-prem estate also removed the hardware refresh cycle, backups and licensing overhead that came with it.