Iceberg Summit 2025
The data team at the largest broadcaster in Europe (+3BN USD / year) had a fragmented ecosystem of cloud (Glue+Hive, Athena, EMR, ECR) and open source solutions (Airflow, Great Expectations): internal adoption of the stack was low, as time to productionize and debug any non-trivial use case was significant. In this talk, we show how the team migrated to open formats over a Python lakehouse in a matter of days, and now enjoys expressive table semantics on S3 backed by Iceberg abstractions. Schema evolution made development easier, time-travel simplified debugging, and pipelines with built-in transactions helped scale adoption 100x throughout the POC. We will conclude by discussing the importance of API design for within-enterprise virality, and highlight the roadmap ahead as Iceberg becomes the center of new AI initiatives in the company.
00:00 - Intro