Case Study: CarMax achieves rapid, modernized daily recommender and omnichannel ML customer experience with Databricks

CarMax - Customer Case Study

CarMax, the nation’s largest used-vehicle retailer and a Fortune 500/S&P 500 company, needed to modernize its recommendation platform to enable an omnichannel customer experience and better empower data scientists. The challenge was to evaluate modern DS/ML tools and architectures, design a scalable batch recommender that uses clicks, views and other signals, and deliver a working system quickly.

Over a four‑month Phase 1 (July–November 2018) the team ran POCs (Dataiku, Databricks, Azure ML Studio, H2O.ai), partnered with vendor experts, and built a metadata‑driven data ingestion framework on a Data Lake with Apache Spark/Azure Databricks. Using two‑week agile sprints they delivered a new daily batch recommender with on‑demand model refresh capability and APIs—deployed in November 2018 to support CarMax’s omnichannel personalization efforts.

CarMax

Todd Dube

Senior Solution Architect