Metaflow makes it quick and easy for data scientists to build and manage real-life machine learning projects. Metaflow can work its magic after the necessary infrastructure for data access, compute, and orchestration has been set up. On the left, you can find material that helps you set up infrastructure for Metaflow in your own environment:
- The easiest way to set up the full stack is to use our CloudFormation template for AWS which uses AWS Batch as the compute layer and AWS Step Functions as the workflow orchestrator.
- Alternatively, you can set up the stack on Kubernetes, using a Kubernetes (EKS) cluster as the compute layer and Argo Workflows as the orchestrator.
- Advanced users may benefit from manual set up instructions or our Terraform template for AWS.
- If you need help setting up the infrastructure, join our slack for support. We can provide detailed guidance and patterns for organizations of all sizes.