MongoDB Atlas + Terraform Spring Boot RAG Application

This repository demonstrates how to build a Retrieval-Augmented Generation (RAG) application using Spring Boot, OpenAI embeddings, and MongoDB Atlas Vector Search. The infrastructure is automated using Terraform to provision and manage MongoDB Atlas resources.

Overview

In this project, we:

• Set up a MongoDB Atlas cluster and configure vector search using Terraform.
• Build a Spring Boot app that generates embeddings with OpenAI, performs vector search on MongoDB Atlas, and handles Retrieval-Augmented Generation (RAG).
• Automate infrastructure provisioning using HashiCorp Terraform.

Features

• Automated Infrastructure: Terraform is used to provision MongoDB Atlas resources including clusters, vector search indices, and access controls.
• RAG Implementation: A Spring Boot application that uses OpenAI to generate embeddings, with MongoDB Atlas vector search to perform semantic searches.
• Document Loading: Upload documents and store their embeddings for use in vector search.
• Querying with Vector Search: Search documents by semantic similarity using a custom /question endpoint.

Prerequisites

To run this project, you'll need:

• Java 21 or higher
• Maven (or Gradle)
• MongoDB Atlas account (with billing configured)
• OpenAI API key
• Terraform installed on your system

Quickstart

1. Clone the repositories

1. Clone this repository for the Terraform setup:

   git clone https://github.com/your-username/mongodb-atlas-rag.git
   cd mongodb-atlas-rag

2. Clone the Spring Boot application repository:

   git clone https://github.com/mongodb-developer/Spring-AI-Rag.git
   cd Spring-AI-Rag

2. Set up MongoDB Atlas with Terraform

1. Create a main.tf and variables.tf as shown in the tutorial.
2. Add environment variables for your MongoDB Atlas credentials:

   export TF_VAR_atlas_org_id="your_atlas_org_id"
   export TF_VAR_public_key="your_public_key"
   export TF_VAR_private_key="your_private_key"
   export TF_VAR_db_username="your_db_username"
   export TF_VAR_db_password="your_db_password"
   export TF_VAR_ip_address="your_ip_address"

3. Initialize and apply Terraform:

   terraform init
   terraform apply

3. Configure the Spring Boot Application

1. Add your MongoDB Atlas URI and OpenAI API Key in the application.properties file:

   spring.data.mongodb.uri=your_mongodb_uri
   spring.ai.openai.api-key=your_openai_api_key

2. Build and run the Spring Boot app:

   mvn clean install
   ./mvnw spring-boot:run

4. Load Documents and Test Queries

• Load documents into the MongoDB vector store:

  curl http://localhost:8080/api/docs/load

• Query the documents using the /question endpoint:

  curl "http://localhost:8080/question?message=How%20to%20analyze%20time-series%20data%20with%20Python%20and%20MongoDB?"

Technologies Used

• Spring Boot: Java-based framework for building REST APIs.
• MongoDB Atlas: An integrated suite of data services (including Atlas Vector Search) centered around a cloud database designed to accelerate and simplify how you build with data. Build faster and build smarter with a developer data platform that helps solve your data challenges. Click here to learn more.
• OpenAI: Generates embeddings for semantic searches.
• Terraform: Automates infrastructure management for MongoDB Atlas.