This repo showcases how to implement a streaming serverless Retrieval Augmented Generation (RAG) architecture.

Customers asked for a way to quickly test RAG capabilities on a small number of documents without managing infrastructure for contextual knowledge and non-parametric memory. In this pattern, I run a RAG workflow in a single Lambda function, so that customers only pay for the infrastructure they use, when they use it.

The responses are streamed using Lambda URL function streaming for a quicker time to first byte and a better user experience. This pattern makes use of Bedrock to calculate embeddings with Amazon Titan Embedding and any Amazon Bedrock chat model as prediction LLM.

I also provide a local pipeline to ingest your PDFs and upload them to S3. There is also an example of how to consume streamed responses using a very basic React application.