A vast and ever-growing volume of Earth observation data is made freely available through initiatives like the ESA Copernicus programme. While these datasets—such as Sentinel-1 and Sentinel-2 imagery—offer immense potential for environmental monitoring, land use assessment, and climate analysis, efficiently accessing and processing them at scale remains a technical challenge. In this tutorial, we demonstrate how STAC (SpatioTemporal Asset Catalog) can serve as a powerful interface for organizing and discovering geospatial assets, and how it can be combined with modern open-source tools such as Dask, Xarray, and Zarr to enable scalable data processing. We will walk participants through workflows that range from querying STAC APIs and loading analysis-ready EO data to performing cloud masking, mosaicking, and time series analysis, all within distributed computing environments (i.e. Dask). The session concludes by showcasing how this pipeline can be extended to machine learning tasks such as land cover classification and deforestation detection, offering participants a robust foundation for building reproducible geospatial analytics workflows.