Case Study: Rue La La achieves scalable, personalized shopping experiences with Amazon Web Services

Rue La La - Customer Case Study

Rue La La, a members-only e-commerce site built around 72-hour flash sales and constantly changing inventory, needed a way to deliver highly personalized product feeds despite massive daily turnover (20–30% new items) and a severe cold-start problem. The data science team set out to build “MyRue,” a collaborative-filtering recommendation engine that could handle large-scale, highly dynamic data while meeting strict latency and availability requirements.

They implemented an ALS-based collaborative-filtering model using Apache Spark MLlib on Databricks and integrated it with AWS services (S3, AWS Batch, DynamoDB, Lambda, API Gateway) for ingestion, storage, and real-time delivery. The system stores 100M+ items in a sparse matrix of ~20B elements, serves recommendations in 26–28 ms even at peak, has run reliably for months with minimal staff, and delivered scalable, cost-efficient personalization that became a competitive advantage.

Rue La La

Stephen Harrison

Data Science Architect