Skip to content
BENIBIX Learn

Kubeflow

Also known as Kubeflow Pipelines

Free tools that run machine learning steps on Kubernetes. Each step gets its own box.

Think of it like

It is like a bus plan for model work. Each step leaves when the one before it is done.

Example

Jack's team already runs its apps on Kubernetes. They use Kubeflow to chain data prep, training and tests, so each night's model is built the same way.

Why it matters

Teams that already use Kubernetes can run model work on the same machines. They do not need a second system.

Definition

Kubeflow is a free set of tools that runs machine learning work on Kubernetes. Its main part, Kubeflow Pipelines, runs each step in its own container. Other parts train a model on many machines at once, or find its best settings. It began as a way to run TensorFlow jobs on Kubernetes. It is open source and free to use.
The bracket under the loop is Kubeflow. It runs each step as a pipeline on Kubernetes.

Common mistake

Picking it for a one-person project. It needs a Kubernetes cluster, which is a lot to run for a small job.

Key takeaways

  1. 1
    Kubernetes first. Runs on the cluster you already have.
  2. 2
    One step, one container. Each step is packed and run apart.
  3. 3
    Free and open. A graduated CNCF project.