Use it when
- You want to set up ML tasks (monitoring, data movement, scaling, versioning, CI/CD integration, and serving) as serverless functions.
- You want to visualize pipelines as DAGs (Direct Acyclic Graphs).
- You want ETL pipelines to extract batch data from multiple sources and run data transformations.
- You want an integrated feature store.
- You want easy built-in functions to test production pipelines locally before deployment.
- You want a marketplace for ML functions to improve your pipelines with minimal engineering efforts.
Watch out
- The self-hosted version requires a Kubernetes cluster. It is possible to run MLRun locally with Docker Desktop or minikube.
- Many functions and services are only available out of the box in managed service offered by Iguazio.
Example stacks
Airflow + MLflow stack
Installation
kubectl create namespace mlrun
helm repo add v3io-stable https://v3io.github.io/helm-charts/stable
helm repo update
kubectl --namespace mlrun create secret docker-registry registry-credentials --docker-server <your-registry-server> --docker-username <your-username> --docker-password <your-password> --docker-email <your-email>
helm --namespace mlrun install mlrun-kit --wait --timeout 960s --set global.registry.url=<registry-url> --set global.registry.secretName=registry-credentials v3io-stable/mlrun-kit