@Artur Dobrogowski Do we we have any python sdk for kubeflow plugin. We are looking for a pythonic way to use the functionalities instead of the CLI being offered.

How would you see that? How would you like to use it?

@Artur Dobrogowski So here is the thing we want our users to edit the Run parameters before they trigger the pipeline from kubeflow UI. We would like to know - 1. Where exactly do we mention these run parameters in the kedro project which can be edited from the kubeflow Ui whenever we are preparing for a new run. Is it the kubeflow.yaml file ?? If you can provide some examples of configuring it , would be really helpful. 2. We also wants to know how can we give unique names to each run for a given pipeline. I see the kubeflow.yaml has these keys experiment_name and run_name . Looks like these fields needs to be edited and we can simply create a new config file before running the command

kedro kubeflow run-once  .

But as I mentioned our users will be launching pipelines from kubeflow UI and they won't be running these commands. Let me know if I am really missing .

I think I'm a bit too rusty with kubeflow right now to address this correctly, I'll ask my colleagues if they can chime in. The 2nd issue can be easily resolved by utilizng the OmegaConf in experiment/run names to append like a date or read something from env variables to append to names in the config. Another thing you can do is provide --params to override any values in the config (kubeflow.yaml) parameters or the run names. I do not remember now with what kubeflow ui shows and allows to edit right now and don't have access to environment with it to check it quickly.

@Vishal Pandey I don’t have where to test it but here’s what I recall. • The parameters that are exposed by kedro-kubeflow after pipeline is deployed are taken form parameters.yaml file, so the standard way kedro handles parameters. • Once you have your pipeline deployed with kedro kubeflow upload-pipeline you should be able to see it in the UI and then manually provide parameter values and specify the run name each time you trigger the new pipeline run.