The document describes Trulia's goal of providing personalized home search experiences and their use of a Lambda architecture to combine batch and real-time processing of user event data at scale. It discusses how they needed an approach to recalculate user profiles from full event data, access them quickly, and add new metrics to past aggregates. Their solution ingests hundreds of millions of daily events, calculates user traits in batch and real-time, and transitions between epochs as batch processing completes to ensure the latest user linkage is always reflected. This allows them to continuously improve personalization by applying new algorithms to past data.