Case Study: Zillow achieves near-real-time, more accurate home-value estimates with Amazon Web Services (AWS)

Zillow Provides Near-Real-Time Home-Value Estimates Using Amazon Kinesis

Zillow Group, owner of the leading U.S. online real-estate brands, runs the Zestimate machine-learning home-valuation tool for more than 100 million homes. As data sources and model complexity grew—tax records, sales, images, MLS data and user inputs—the company’s on-premises framework couldn’t scale, and Zestimates could take a day or more to compute, leaving customers with stale information.

Zillow moved ingestion to Amazon Kinesis, built a centralized data lake on Amazon S3, and runs distributed machine learning with Apache Spark on Amazon EMR. This architecture lets Zillow process streaming data and execute massively parallel jobs so Zestimates are computed in seconds to hours instead of a day, improving accuracy and enabling scalable analytics, personalization, and ad targeting across the business.

Zillow

Jasjeet Thind

Vice President of Data Science and Engineering