Charmed Kubeflow allows data scientists and engineers to rapidly develop sophisticated machine learning (ML) and artificial intelligence (AI) models, advanced analytics, and data engineering pipelines at scale by scheduling their jobs and workflows on Kubernetes clusters.
During the research and development phase though, data scientists and engineers need a lab environment where they can develop their models, data preparation and feature engineering scripts, and other items like visualisations and analyses. Whilst most typically data scientists want to use JupyterLab for this, there are some other options available on Charmed Kubeflow, most notably VSCode.
What is VSCode?
Launching VSCode on Charmed Kubeflow
Let’s launch VSCode now. From the Kubeflow Dashboard, navigate to “Notebooks” and click “New Notebook”.
In the next screen, enter a name for the notebook server. You could call it “vscode”. Check the box labelled “Custom Image” and in the field below, enter the following image name:
Choose how much RAM, how many CPU cores and any GPUs that you want to assign, then scroll down and click “Launch”.
Charmed Kubeflow will now start the VSCode server for you, and after a few moments a green tick will appear, signalling that you can connect to the IDE. Click “Connect” and a new browser tab will open.
In the new browser tab, you will see VSCode running, ready to go. You can create new files, install plugins, clone git repositories and most of the other things you would expect to be able to do.