Arvid Vinther Normann

Novo Nordisk transformed how vendor-sourced and vendor-facing clinical trial data is ingested, operationalized and exposed by building an API layer on the Databricks lakehouse powered by Lakebase. By replacing fragmented, self-hosted Postgres silos with Lakebase, a Postgres engine integrated with the lakehouse, we now make data available 30x faster than with legacy self-hosted Postgres. This session walks through the end-to-end architecture and migration. We’ll talk about why Lakebase was chosen—horizontal scale and simplified operations—and how these capabilities reduce friction and speed onboarding. We’ll further expand on the blueprints we employed for Lakebase development, including governance and deployment patterns that can be adopted across an organization, and elaborate on the lessons learned along the way. Attendees will learn how we modernized ingestion of structured and unstructured data to achieve consistent, auditable data in a regulated environment.