Case Study: Snorkel achieves 96% faster agent evaluation cycles with Daytona

Snorkel speeds agent evaluation cycles by 96% with Daytona

Snorkel, a company specializing in AI data infrastructure, faced a challenge scaling its agent simulations. They needed to provision thousands of isolated, configurable sandboxes to run parallel evaluations without workloads interfering. Building this complex runtime infrastructure in-house would have diverted focus from their core AI initiatives, leading them to partner with Daytona for its runtime platform.

By integrating Daytona, Snorkel gained an on-demand solution for provisioning isolated sandboxes via a simple API. This allowed them to run massive numbers of simulations in parallel with customized compute resources. The solution from Daytona resulted in a 96% acceleration of evaluation cycles and the provisioning of over 1 million sandboxes, enabling Snorkel to ship reliable, frontier-grade AI training datasets much faster.

Snorkel