@Joe Mesterhazy Thanks! It sounds like your use case mirrors my own. I wanted to run CPU-bound tasks using Python's multiprocessing builtins without needing to pull in Dask or Ray. Not that they aren't great - just, like you said, they feel like overkill for just running things locally.
There are still some bugs to work out that only appear under certain circumstances. I just released a new beta that fixes one of them! If you have a chance, please feel free to install v0.2.0b4 and see if it handles your workloads. If you run into any problems, it'd be great if you could open a GitHub issue.