Production Workflows

Build systems around highly available workflows, SLA guaranteed.

One-click production deployment

Any data developer can develop and test production-grade data and ML workflows locally and deploy them in production with a single click - no changes in the code required.

Unify data and ML, AI workflows on a unified platform, following the best practices developed at Netflix and other modern data-driven organizations.

Hardening flows for production

Configure retry strategies, fallbacks, and timeouts to harden your flows against surprises. Observe failures and fix issues quickly locally.

Use Metaflow’s built-in dependency management to create stable execution environments or bring your own Docker images.

Integrate to other flows and systems

Flows are not islands. They can react to events and data from surrounding systems and other flows.

The platform comes with a highly available event-bus that allows you compose sophisticated, reactive data and machine learning systems.

Test production deployments in the Metaflow sandbox

Test production deployments in the Metaflow sandbox