​Apache Iceberg™ Europe Meetup - Berlin

​Join us for the Apache Iceberg™ Europe Meetup! Our event is hosted in Berlin co-hosted by Google Cloud, Databricks and Vakamo.

​🎟️ When registering, please select one of the two ticket types:

• ​In-Person Ticket: Join us on-site! Your name will be used to pre-register for venue access.

• ​Remote-Only Ticket: Can’t make it in person? No worries—register to join the live stream, receive event recordings, and stay connected with the community.

​Agenda

​5:00 pm – Registration & Networking
6:00 pm –Meetup starts

​🎙️ Eduard Tudenhoefner (Databricks): What's coming with Iceberg v4?

​🎙️Maximilian Michels (Apple): Cheap writes, fast reads: converting equality deletes to deletion vectors

​🎙️ Jan Kaul (Vakamo): Iceberg Materialized Views: Portable and Engine-Agnostic

​Network break

​🎙️ Vladislav Sidorovich (Google Cloud): Multi-cloud data access with Gemini and Apache Iceberg

​🎙️ Philipp Fischbeck: (Snowflake) : Multimodal Iceberg: File and Vector Data Types in Iceberg V4 Tables

​20:30 More Networking

​21:00 Meetup closes

​Livestream

​How to Get to the Venue

​

​Building Access

​Goolge Cloud

​Tucholskystraße 2

​10117 Berlin, Germany

​🪪 ID Requirements

​please bring you ID with you

​Presentations & Speakers

​​​​​​​​​​​​​🌟What's coming with Iceberg v4?

​This talks gives an overview of what features and changes are being introduced with the Iceberg v4 spec, what work has already landed in the Java implementation, and how to get involved.

​Eduard is Staff Software Engineer / Apache Iceberg committer and PMC member

​​​​​​​​​​🌟 Cheap writes, fast reads: converting equality deletes to deletion vectors

​Engines like Flink can stream upserts and CDC changes into Apache Iceberg, but every update writes an equality delete. Equality deletes are cheap for the writer and expensive for everyone else: each query has to match them against the data, and the cost grows with every commit until the table is rewritten.

​ConvertEqualityDeletes, a Flink maintenance task in Apache Iceberg 1.12.0, removes that cost without changing the writer. After each commit, it looks up the position of every deleted row in a primary-key index kept in Flink state, and replaces the equality deletes with deletion vectors. Readers then skip deleted rows by position.

​In this talk, we'll cover how the conversion works, how it handles out-of-order upserts, compaction, and failures, and how to enable it with one line on the IcebergSink. We'll also look at why the community used it as the case for dropping equality deletes from v4, and what comes next: sharing the index across engines and writing deletion vectors directly.

​Max is a software engineer at Apple who loves open-source, distributed systems, and stream processing. Max previously worked on large-scale data processing tools with Google, Lyft, and Splunk. Max is a PMC member at the Apache Flink and Apache Beam, and an Apache Iceberg committer.

​​​​​​​​​​​🌟 Iceberg Materialized Views: Portable and Engine-Agnostic

​Materialized views are a proven way to speed up queries and cut compute costs. But an MV's semantics, metadata and refresh logic usually belong to a single engine, so it can't be shared or maintained anywhere else. Apache Iceberg changes that. Its table and view specs, snapshot isolation and versioned metadata give MVs a foundation that isn't tied to one engine. The community has worked on a MV spec that extends table and view metadata to track dependencies and refresh state. With that metadata, any engine can maintain, validate and read the same MV. This talk covers the key design choices behind the spec, what it makes possible today and where Iceberg MVs go next.

​Jan Kaul recently joined Vakamo, the company behind Lakekeeper, an open-source Apache Iceberg REST catalog written in Rust. He is the initiator of iceberg-rust, the Rust implementation of Apache Iceberg. He also built Frostbow, an in-process analytical query engine for Iceberg tables in object storage. His work focuses on materialized views and incremental view maintenance. Jan holds a PhD in mechanical engineering from TU Wien and describes himself as an engineer turned programmer.

​​​​​​​​​​​​​🌟 Multi-cloud data access with Gemini and Apache Iceberg

​Discover how to unlock multi-engine analytics by running native BigQuery SQL (DML, DDL, and AI functions) directly over Apache Iceberg tables — and explore the architecture behind it.
We'll walk the architecture end to end: how the Lakehouse Iceberg REST Catalog keeps metadata transactionally consistent while your data stays in open Iceberg format in your own Cloud Storage buckets, and how query engines read and write the same tables without vendor lock-in.

​Vladislav Sidorovich, SWE at Google Cloud

​​​​​​​​​​​​​🌟 Multimodal Iceberg: File and Vector Data Types in Iceberg V4 Tables

​The Iceberg table format version 4 introduces two new data types: file and vector. In this session, we provide an overview into how these new types work. We discuss the changes that were necessary in Parquet, the trade-offs discussed in the Iceberg and Parquet community around the types, and how engines like Snowflake can enable new workloads in the AI/ML field on interoperable Iceberg tables.

​Philipp Fischbeck is a software engineer on the SQL data types team at Snowflake, where he works on making complex data shapes convenient and efficient to store and query. His work focuses on Snowflake's core SQL engine capabilities and its integration with open formats like Apache Iceberg and Apache Parquet. Most recently, he contributed to the community’s efforts around the vector data type.