The talk at the FOSDEM conference was about change data capture (CDC) and stream processing using Apache Flink. The speaker discussed the different building blocks of stream processing and how Flink handles them. They also highlighted the unique features of Flink, such as its ability to create snapshots of the entire stream, its scalability and parallelism, and its support for various use cases and connectors. The talk also covered the different APIs of Flink, including the DataStream API, Table API, and Stateful Functions. The speaker explained the stream-table duality concept and the different types of change logs in Flink. They also discussed the concepts of retracting and upserting, and how Flink handles transactions and running times. The talk concluded by mentioning other sub-projects of Flink, such as Flink SQL, table store, and the capabilities for handling big state and ensuring exactly once semantics.