Case Study: Anastasia accelerates ML processing 9x and cuts costs 87% with Anyscale and Ray

How Anastasia accelerated their ML processes 9x with Ray and Anyscale

Anastasia, an AI-powered forecasting platform, needed a more scalable and cost-effective way to run demand prediction and model training for customers forecasting everything from spare parts to product demand. Their existing Python and AWS Batch approaches hit limits in vertical scaling, added infrastructure complexity, and made horizontal scaling and hyperparameter tuning difficult.

Using Ray and Anyscale, Anastasia redesigned its ML pipeline to run as a true cluster with autoscaling, Ray Tune for distributed hyperparameter search, and easier training across many time series models. The result was a 9x faster workflow and an 87% reduction in cost compared with their AWS Batch implementation, while Anyscale further improved development speed, logging, governance, and production testing.

Anastasia

Juan Roberto Honorato

AI Tech Lead worker