Hybrid search over California embeddings with Modal, MongoDB, and Clay

A hybrid satellite-image search application demonstrates how Modal’s serverless infrastructure and MongoDB Atlas can combine data APIs, foundation models, and geospatial search to explore California by timestamp, location, and image similarity. The system regularly retrieves new Sentinel satellite imagery through an API, stores its JSON metadata and geographic areas of interest in MongoDB Atlas, identifies records lacking embeddings, and sends image regions to a Clay v1 satellite-imagery model running on A10 GPUs through Modal. Generated embeddings are stored alongside the source data, enabling an Alpine JS and FastAPI-based interface to issue hybrid MongoDB Atlas queries that blend geolocation, vector embeddings, and time filters. The project is presented as an example of a broader application pattern in which Modal supports scheduled tasks, API services, and GPU inference while Atlas supplies flexible JSON storage, vector, text, and geospatial search, along with globally scalable data-management capabilities.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.