Medallion Architecture in Practice: Best Practices, Pitfalls and the Impact of AI

Medallion architecture is a popular pattern for progressively refining data, but successful implementation takes more than setting up Bronze, Silver and Gold. How do you define layer boundaries? Where do data quality, modelling and ownership belong? And how do you avoid unnecessary copies, complexity and costs? Drawing on his book Building Medallion Architectures and experience across numerous enterprises, Piethein Strengholt shares practical best practices, do’s and don’ts, and a concrete implementation roadmap. He also examines Medallion through the lens of AI: what still holds, what needs adapting, and where do unstructured data, embeddings and retrieval fit? A practical session for anyone looking to apply Medallion effectively and think beyond the familiar three-layer model.

• From Bronze to Gold: clear responsibilities and criteria for moving data between layers
• Data quality, data modelling and ownership: which decisions belong where?
• Enterprise do’s and don’ts: common pitfalls and how to avoid them
• A practical roadmap: from an initial implementation to a manageable architecture at scale
• Medallion in the AI era: the role of unstructured data, embeddings and retrieval
• Looking ahead: where Medallion continues to add value and where complementary architecture patterns are needed.