Context I am a machine learning engineer, but one of my team's goal is to build a sort of internal developer platform that can provide VS Code as a service with a customised python environment (will be used mostly for python ETL and model training/serving scripts). What I Tried Experimenting with remote tunnel option on VSCode and got it working in my Docker container after passing through a few interactive steps (including sign in to GitHub). The docker container gives me a URL containing a sort of tunnel-id and I can access VS code (running inside the container) over the internet. What I Want Something similar to this, but best if with some of the following characteristics Exposed as a service on a Kubernetes cluster within my company's intranet Has a prebuilt specific python environment, i.e. having specific versions of libraries (e.g. tensorflow installed) The environment is set up to access my company's BigQuery data warehouse and other Google cloud resources Some auto-scalability, so that when, suppose 10 users are logging in, the service can still bear the load, and we can recover from crash I may not be articulating every requirement clearly, as I have to figure out a lot as I go along but the above should give some idea of what I am looking for. Is that even possible, using open source (and basic Google cloud solutions like GKE) as much as possible? What kind of documentations should I look at? Any pointer or resource would be appreciated. Continue reading...
