Matthew Schumpert

Your Iceberg lakehouse gives you open access to structured, historical data via SQL. But what about events that arrived in the last five seconds — data still in motion? Most data teams have a blind spot: they can query history, but not the present. Skilled engineers can build Flink or Spark jobs to access Kafka/Redpanda, but mere mortals can’t. Some events arrive late. Others simply don’t exist in the lakehouse. These gaps matter: delayed fraud reports, stale ML features, AI agents reasoning on yesterday's context and making bad decisions. Bad agent! In this session, Redpanda's Matt Schumpert explores how streaming data and Iceberg tables can be unified in a single SQL query. Streams are just tables in Redpanda SQL. You'll see how a Postgres-compatible query layer bridges live events and Iceberg tables in S3 transparently, and what that means for your analytics and AI stack.