RAG Pipeline with Node.js and Oracle Database

A RAG pipeline involves acquiring data from a data source, online or offline, chunked and stored in a vector database such as the Oracle Database 23ai vector database and then a subset of this data is retrieved contextual to a question or user query which is then further fed to a LLM to generate a meaningful response.

Node.js can be used to develop and host powerful AI applications that can utilize frameworks such as Next.js , MUI, React to design and develop user interfaces.

EmbedJs is an open source javascript framework which provides excellent tooling for creating RAG pipelines that can serve as backend hosted on Node.js.

The example at https://github.com/lake-of-dreams/oracledb-embedjs utilizes the features provided by EmbedJs as a RAG pipeline framework and Oracle Database as the Vector database, with the help of node-oracledb driver and the Ollama engine to run on-demand models.

Whats more ? You don't need anything apart from either docker or podman and a Node.js installation to run the example as we use the docker-compose library to manage database and Ollama containers !

Please feel free to try and provide your feedback.