We get asked a lot about dense array geospatial workloads where data is stored in S3 in zarr format and processed with xarray. Lots of people like xarray-cupy and want example of how to deploy Dask clusters that use xarray, and how we can use CuPy and kvikio to accelerate workflows.