> When I compile code, I want to fire up 1000 serverless container and compile tiny parts of my code in parallel.

Or maybe we should make better use of GPU and other local compute.

https://github.com/StanfordSNR/gg