RAG App (Retrieval-Augmented Generation)

This repository contains a full-stack Retrieval-Augmented Generation (RAG) application. The backend is built with FastAPI and integrates with ChromaDB and Ollama for LLM inference and vector storage. The frontend is built with Angular and provides a user-friendly interface for document upload, querying, and viewing responses.

---

Table of Contents

• Features
• Project Structure
• Prerequisites
• Setup Instructions
• Running the Application
• API Endpoints
• Sample Inputs
• Advanced RAG Techniques
• Available Scripts
• Implementation Overview

---

Features

• Document Upload: Upload PDF or text files for ingestion and retrieval.
• Query Submission: Submit queries and receive retrieval-augmented responses.
• Advanced Retrieval: Supports hybrid, reranking, and semantic search strategies.
• Interactive UI: Angular-based frontend for seamless user experience.
• API Documentation: Swagger UI for backend API exploration.

---

Project Structure

.
├── backend/
│   ├── README.md
│   ├── requirements.txt
│   ├── chromadb/
│   ├── sample_inputs/
│   └── src/
│       ├── main.py
│       ├── api/
│       ├── models/
│       ├── services/
│       └── utils/
└── frontend/
    ├── README.md
    ├── angular.json
    ├── package.json
    ├── tsconfig.json
    └── src/
        ├── index.html
        ├── main.ts
        ├── polyfills.ts
        ├── styles.css
        └── app/

---

Prerequisites

Backend

• Python 3.8+
• pip
• virtualenv (recommended)
• Ollama (for local LLMs)
• ChromaDB (included as files)

Frontend

• Node.js (v14+)
• npm
• Angular CLI (v13.x)

---

Setup Instructions

Backend

1. Clone the Repository

   git clone <repository-url>
   cd backend

2. Create and Activate a Virtual Environment

   python3 -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install Dependencies

   pip install -r requirements.txt

4. Install Ollama

   • macOS: brew install ollama or download from https://ollama.com/download
   • Windows: Download and install from https://ollama.com/download
   • Start Ollama: ollama serve
   • Pull a model: ollama pull llama3

5. Prepare ChromaDB

   • The chromadb/ directory is pre-populated. No manual setup required.

6. (Optional) Add Environment Variables

   • Create a .env file in backend/ if needed.

---

Frontend

1. Clone the Repository

   cd frontend

2. Install Dependencies

   npm install

3. Install Angular CLI (if not already)

   npm install -g @angular/cli@13

---

Running the Application

Backend

1. Start Ollama

   ollama serve

2. Start FastAPI Server

   cd backend/src
   uvicorn src.main:app --reload

   • API docs: http://127.0.0.1:8000/docs

Frontend

1. Start Angular Development Server

   cd frontend
   npm start

   or

   ng serve

   • App available at http://localhost:4200

---

API Endpoints

• POST /upload: Upload a document (PDF or text) for ingestion.
• POST /query: Submit a query and receive a retrieval-augmented response.
• GET /health: Health check endpoint.

See Swagger UI for detailed schemas.

---

Sample Inputs

The backend/sample_inputs/ directory contains:

• meditations.pdf: Example PDF for ingestion.
• sample_query.txt: Example query.
• sample.json: Example JSON input.
• sample.txt: Example plain text file.

---

Advanced RAG Techniques

When submitting a query, users can select the retrieval strategy:

• Basic: Standard retrieval-augmented generation.
• Hybrid Search: Combines keyword and semantic search.
• Reranking: Retrieves candidates and reranks them using a secondary model.
• Semantic Search: Uses embeddings for retrieval.

The selected technique is sent to the backend as a technique parameter in the /query endpoint.

---

Available Scripts

Backend

• Start server: See above.
• Install dependencies: pip install -r requirements.txt

Frontend

• npm start / ng serve: Start dev server.
• npm run build / ng build: Build for production.
• npm test / ng test: Run unit tests.
• npm run lint / ng lint: Lint codebase.
• npm run e2e / ng e2e: Run end-to-end tests.

---

Implementation Overview

• Backend: FastAPI app with endpoints for document upload, query, and health check. Uses ChromaDB for vector storage and Ollama for LLM inference.
• Frontend: Angular app with components for upload, query, and results. Uses a dedicated service for API integration.
• Component-Based UI: Upload, query, and results display are organized as reusable Angular components.

---