Case Study: Turo achieves faster ML feature delivery and predictable low-latency performance with Chalk

Turo ships new production features in 1 week with Chalk

Turo, the world's largest car-sharing marketplace, faced a challenge with fragmented and inconsistent machine learning feature pipelines. This lack of a unified feature store slowed their time to production for new features, creating a bottleneck for their search, pricing, and risk models. They partnered with vendor Chalk to implement its compute-first feature store to address these issues.

By adopting Chalk, Turo established a centralized, self-serve platform for feature delivery. This solution standardized workflows and reduced dependencies on other teams, allowing the ML engineering team to ship new production features in one week instead of two to three. Chalk provided predictable, low-latency performance, enabling faster iteration and creating a scalable foundation for real-time ML.

Turo