Developed and delivered two end-to-end Jupyter notebooks for the IBM/terratorch repository, enabling users to generate embedding vectors from Sentinel-2 RGB imagery using the TerraMind deep learning model. The work focused on integrating geospatial data with machine learning workflows, providing reproducible templates that streamline model setup, image processing, and embedding retrieval. Leveraging Python, PyTorch, and geospatial analysis techniques, the notebooks were designed to facilitate rapid experimentation and onboarding for geospatial deep learning projects. This contribution enhanced the repository’s value for client demonstrations and practical use cases, supporting business needs through clear, ready-to-run examples without addressing major bug fixes during the period.