Application
Automation & IntegrationFine-tuning & Training
Kubeflow
1Machine 2learning 3toolkit 4built 5for 6Kubernetes
the six Ws · specification
W1
Who
Platform and ML engineers running scalable machine learning pipelines on Kubernetes clusters.
W2
What
A Kubernetes native toolkit for orchestrating ML pipelines, distributed training, hyperparameter tuning and model serving.
W3
Where
Self-hosted, deployed onto any Kubernetes cluster, on premises or in any cloud.
W4
When
Launched by Google in 2018, now a CNCF project with continued releases through 2026.
W5
Why
Brings Kubernetes-grade scalability and portability to end-to-end machine learning workflows and pipelines.
W6
With
Requires a running Kubernetes cluster plus components such as Argo Workflows, Katib and KServe.
W7
Watch
Open source under Apache-2.0, self-hosted, no vendor account required, governed as a CNCF project, over 15k GitHub stars.
alternatives