Mark is back with another recipe where he explains the process of rolling up segments in real-time tables in Apache Pinot. This process aims to reduce the amount of space that data occupies by aggregating it up to the nearest minute, hour, or even day. He uses an example involving product purchases streamed in Kafka, which are aggregated according to parameters such as minimum stock, sum of quantity, maximum price, and sum of sales amount. Needham also demonstrates how these rollups are configured in Apache Pinot, and visualized via the Pinot UI, while explaining the concept of task types, ingestion configuration, segment names, and the influence of time intervals on aggregation.