PopChoice

This repository is a movie recommendation system demonstrating the integration between CloudFlare Workers, CloudFlare AI Gateway, CloudFlare Pages, Supabase/PostgreSQL with PGVector extension, and OpenAI API for semantic search capabilities.

Architecture

The application consists of three main components:

• A frontend web application built with vanilla JavaScript
• An OpenAI embedding worker for generating vector embeddings
• A Supabase/PostgreSQL worker for performing vector similarity search
• A scheduled worker that performs a health check query towards Postgresql once every day; to avoid the project begin suspended in Supabase.

Key Features

• Vector embeddings generation using OpenAI's text-embedding-ada-002 model
• Vector similarity search using PostgreSQL with PGVector extension
• Scheduled ingestion of movie data using CloudFlare Workers
• CORS-enabled API endpoints for cross-origin requests