Have you considered looking at data science use cases for this (particularly with Python)? I've spent a lot of time tweaking containers meant for data science and one issue I run into a lot is that since you actually spend a lot of time developing inside the containers you want access to a lot of developer tools, but you end up with enormous images as a result. I could see using a tool like this to have one main developer container for working from that has all of your tooling and then using more streamlined containers for individual projects that have only the dependencies you need.