At Descartes Labs, we use the commercial cloud to run global-scale machine learning applications over satellite imagery. Our core software stack builds upon many of the OSGeo projects including GDAL and Mapserver, as well as the rich ecosystem of Python libraries. Internally we have processed over 5 Petabytes of public and private satellite imagery, including the full data corpus from the Landsat and Copernicus missions. By combining the above open-source tools with a FUSE-based filesystem for cloud storage, we have enabled a scalable compute platform that has demonstrated reading over 200 GB/s of satellite imagery into cloud compute nodes. In particular, we have generated global 15m Landsat-8, 20m Sentinel-1, and 10m Sentinel-2 composites, using over 10,000 CPUs. We recently created a public open-source python client library that can be used to query and access imagery from within our platform, and made this platform available to researchers for non-commercial projects. In this talk we will describe how researchers can get started on the Descartes Labs Platform, and demonstrate examples in crop yield analysis and land use/land cover classification.