← all repositories

RafalWilinski/cloudflare-rag

A fullstack application enabling users to chat with PDFs using retrieval-augmented generation on Cloudflare's edge infrastructure.

602 stars TypeScript RAG · Search
cloudflare-rag
Velocity · 7d
+0.9
★ / day
Trend
steady
star history

This application allows users to upload PDFs and query them conversationally using LLMs. It implements hybrid RAG combining full-text search on Cloudflare D1 with vector similarity search on Vectorize, supports multiple LLM providers (OpenAI, Groq, Anthropic) via AI Gateway with fallback mechanisms, and streams responses to the frontend using Server-Sent Events. OCR processing runs inside Cloudflare Workers to extract text from uploaded documents.

heatdrop uses Google Analytics to see which pages get read — nothing else. Your call. How we handle data.