This document discusses using AWS services like S3, EMR, and Hadoop to process large amounts of customer data and personalize digital advertising for a large retail client. Some key details: - The client imports 200+ GB of transaction data daily from their existing system to S3 storage on AWS. - An EMR Hadoop cluster with 100 machines is created on demand to process the data using Cascading and segment customers into personalized groups. - This automated process on AWS was able to reduce processing time from over 2 days to just 8 hours and increased the client's return on ad spend by 500%. - The elastic AWS infrastructure allows the client to dynamically scale processing as needed based on data