Description
Large multidimensional arrays can be read and written efficiently across storage backends. Researchers and data engineers can work with chunked scientific datasets, model data, and cloud-backed arrays without loading everything into memory. It may access local or remote storage depending on configuration.