Installation guide

Kedro setup

First, you need to install base Kedro package in <17.0 version

Kedro 17.0 is supported by kedro-kubeflow, but not by kedro-mlflow yet, so the latest version from 0.16 family is recommended.

$ pip install 'kedro<0.17'

Plugin installation

Install from PyPI

You can install kedro-kubeflow plugin from PyPi with pip:

pip install --upgrade kedro-kubeflow

Install from sources

You may want to install the develop branch which has unreleased features:

pip install git+

Available commands

You can check available commands by going into project directory and runnning:

$ kedro kubeflow
Usage: kedro kubeflow [OPTIONS] COMMAND [ARGS]...

  Interact with Kubeflow Pipelines

  -e, --env TEXT  Environment to use.
  -h, --help      Show this message and exit.

  compile          Translates Kedro pipeline into YAML file with Kubeflow...
  init             Initializes configuration for the plugin
  list-pipelines   List deployed pipeline definitions
  run-once         Deploy pipeline as a single run within given experiment.
  schedule         Schedules recurring execution of latest version of the...
  ui               Open Kubeflow Pipelines UI in new browser tab
  upload-pipeline  Uploads pipeline to Kubeflow server


init command takes one argument (that is the kubeflow pipelines root url) and generates sample configuration file in conf/base/kubeflow.yaml. The YAML file content is described in the Configuration section.


ui command opens a web browser pointing to the currently configured Kubeflow Pipelines UI. It’s super useful for debugging, especially while working on multiple Kubeflow installations.


list-pipelines uses Kubeflow Pipelines to retrieve all registered pipelines


compile transforms Kedro pipeline into Argo workflow (Argo is the engine that powers Kubeflow Pipelines). The resulting yaml file can be uploaded to Kubeflow Pipelines via web UI.


upload-pipeline compiles the pipeline and uploads it as a new pipeline version. The pipeline name is equal to the project name for simplicity.


schedule creates recurring run of the previously uploaded pipeline. The cron expression (required parameter) is used to define at what schedule the pipeline should run.


run-once is all-in-one command to compile the pipeline and run it in the Kubeflow environment.