Kubeflow is an end-to-end Machine Learning (ML) platform for Kubernetes, it provides components for each stage in the ML lifecycle, from exploration through to training and deployment. Operators can choose what is best for their users, there is no requirement to deploy every component. To read more about the components and architecture of Kubeflow, please see the Kubeflow Architecture page.
There are two pathways to get up and running with Kubeflow, you may either:
Install a packaged Kubeflow distribution
See the table below for a list of options and links to documentation:
|Kubeflow on AWS||Amazon Web Services (AWS)||Amazon Elastic Kubernetes Service (EKS)||1.3||Docs|
|Kubeflow on Azure||Microsoft Azure||Azure Kubernetes Service (AKS)||1.2||Docs|
|Kubeflow on Google Cloud||Google Cloud||Google Kubernetes Engine (GKE)||1.5.0||Docs|
|Kubeflow on IBM Cloud||IBM Cloud||IBM Cloud Kubernetes Service (IKS)||1.5||Docs||External Website|
|Kubeflow on Nutanix||Nutanix||Nutanix Karbon||1.5.0||Docs|
|Kubeflow on OpenShift||Red Hat||OpenShift||1.3||Docs||External Website|
|Argoflow||Argoflow Community||Conformant Kubernetes||1.3||N/A||External Website|
|Arrikto Kubeflow as a Service||Arrikto||Fully Managed||1.4||N/A||External Website|
|Arrikto Enterprise Kubeflow||Arrikto||EKS, AKS, GKE||1.4||Docs||External Website|
|Charmed Kubeflow||Canonical||Conformant Kubernetes||1.4||Docs||External Website|
Install the Kubeflow Manifests manually
The Manifests Working Group is responsible for aggregating the authoritative manifests of each official Kubeflow component. While these manifests are intended to be the base of packaged distributions, advanced users may choose to install them directly by following these instructions.