feat: added dev container file for fast onboarding

[nkkko]

I have added a devcontainer.json file which pulls a python image from MSFT registry and installs numpy plus adds python extensions for faster and easier onboarding of students.

This works with local VS Code, Codespaces or Daytona.io.

[karpathy]

I haven't actually used this extensively in the past. The links in the comments are not super helpful or illuminating. How do you use this personally

[nkkko]

Hey Andrej, yeah, I use this devcontainer setup as my go-to dev environment. It's essentially a throwaway laptop. No need to mess with Python virtual envs or worry about system-level conflicts. Also works well for quick PR reviews when you need to run the whole env (e.g. Astro web or docs).

You know how in your courses, you'd probably spend the first session just getting everyone's environments sorted? This completely eliminates that. I use it when I teach my Data Science students, everyone starts from the same spot, and gets an identical, pre-config env.

It's not just about saving time (though it does that too). It's about removing this whole layer of cognitive overhead for learners. They can dive straight into the code, the algorithms, the actual interesting stuff, without fighting their tools.

Think of it like how you use notebooks to make concepts immediately tangible. This does the same thing, but for the entire development environment.

I've found it particularly powerful for students who are just starting out. It lowers the barrier to entry significantly. No more "works on my machine" issues.