I wanted to see how the idea of a "time-sharing" computer from the 1970s could be applied to modern computing tasks. The API for GitHub and GitHub Actions was ideal for being able to submit bash scripts and have them execute in a VM, including secrets management and getting binary results back to the local machine like a YouTube video from youtubedl or a custom Linux Kernel built on speedy hardware. Because GitHub's all about AI now, I've also included an example of how to run llama2 to answer a bunch of questions using a pre-trained model from Hugging Face.