FinSight RAG

AI-powered financial intelligence system for analyzing earnings reports and 10-K filings using Retrieval-Augmented Generation (RAG).

Live Demo: https://finsight-rag.vercel.app

Overview

FinSight RAG is a full-stack application that allows users to upload financial documents and query them using natural language. The system retrieves relevant information from documents and generates context-aware answers.

The project is designed to go beyond basic RAG implementations by focusing on structured financial data and real-world document analysis.

Features

• Upload and process financial PDF documents
• Ask natural language questions about document content
• Semantic search using vector embeddings
• FastAPI backend for high-performance API handling
• React frontend with modern UI
• Modular architecture for scalability

Architecture

Frontend (React + Vite)
→ API Layer (FastAPI)
→ Embedding and Retrieval (ChromaDB)
→ Language Model (OpenAI)

Tech Stack

Backend:

• FastAPI
• Python
• ChromaDB
• OpenAI API
• Uvicorn

Frontend:

• React (Vite)
• Tailwind CSS
• Axios

Project Structure

finsight-rag/
├── src/                 # Backend (FastAPI)
├── finsight-ui/         # Frontend (React)
├── chroma_db/           # Vector database
├── data/                # Uploaded files
├── requirements.txt
├── .env
└── README.md

Setup and Installation

Clone the repository

git clone https://github.com/tarekjundi10/finsight-rag.git
cd finsight-rag

Backend Setup

python -m venv venv
.\venv\Scripts\Activate.ps1
pip install -r requirements.txt

Create a .env file:

OPENAI_API_KEY=your_api_key_here

Run backend:

python -m uvicorn src.api.main:app --reload

Backend: http://127.0.0.1:8000

Docs: http://127.0.0.1:8000/docs

Frontend Setup

cd finsight-ui
npm install
npm run dev

Frontend: http://localhost:5173

Usage

1. Upload a financial document (e.g., 10-K report)
2. Click "Ingest Documents"
3. Ask questions about the document

Example queries:

• What is the total revenue?
• What are the key risk factors?
• How did operating income change year over year?

Known Limitations

• Requires a valid OpenAI API key
• Performance depends on document size
• No authentication system implemented

Future Improvements

• Multi-document comparison
• Financial table parsing
• Authentication system
• Docker deployment
• Performance optimization

Project Value

This project demonstrates:

• End-to-end RAG system design
• Full-stack development
• AI integration in real applications
• Handling unstructured financial data