Our geospatial analysis research applies AI and computer vision to understand and analyze geographic data at multiple scales. Recent work includes Earth Embeddings towards AI-centric representations of our planet, retrieval-augmented neural fields for multi-resolution geo-embeddings (RANGE), and active geospatial search frameworks. We also develop methods for weakly supervised building segmentation from aerial images, estimating displaced populations from overhead imagery, and geospatial active search for efficient resource allocation. Our research spans from local urban analysis to global-scale environmental monitoring, leveraging satellite imagery, aerial photography, and ground-level images to build comprehensive geospatial understanding.