TechPrufer
Get started
Modal
Modal logoModal · Feature trail

Hybrid Search with Embeddings

1 post · Sep 2024

What this trail covers

Modal's platform supports building hybrid search applications by integrating with data platforms like MongoDB Atlas and foundation models for generating embeddings. This enables complex search queries combining text, geolocation, timestamps, and vector embeddings derived from satellite imagery, demonstrating a new capability for data-driven AI applications. This post details a specific implementation using Sentinel satellite imagery, Clay v1 foundation model, and MongoDB Atlas for hybrid search over California.

All Modal trails
  1. Hybrid search over California embeddings with Modal, MongoDB, and ClayThis post details a specific implementation of hybrid search using Modal, MongoDB Atlas, and the Clay v1 foundation model for satellite imagery. It outlines the workflow from data ingestion (Sentinel satellite images) to embedding generation (using a serverless GPU service with Clay v1) and hybrid search queries (combining geo-location, image embeddings, and timestamps) executed against MongoDB Atlas. The frontend is a simple Alpine JS client served via FastAPI. The post highlights the integration of these components to create a functional hybrid search application.