This project reflects many insightful discussions with the data science + python community, in particular: developers of Spark and PySpark (esp. Matei Zaharia, Xiangrui Meng, and Josh Rosen), developers of spylearn (esp. Olivier Grisel, Sandy Ryza, and Uri Laserson), developers of sparkit-learn, developers of Dask (esp. Matthew Rocklin), developers of x-ray (esp. Stephan Hoyer). Some of the ideas here for using Spark with ndarrays originated in Thunder (thanks to Jascha Swisher, Michael Broxton, Logan Grosenick, Noah Young, Ben Poole), and special thanks to Stephan Hoyer for the motivation to generalize these concepts.