What if you could write standard numpy and pandas code but have it run on a distributed computing grid for incredible parallel processing right from Python? How about just splitting it across multiprocessing to escape the limitations of the GIL on your local machine? That's what Dask was built to do.