AI Document Analyzer

An AI-powered multi-space document analysis platform where users can upload documents, ask questions about them, and receive context-aware answers with source citations.

The system uses Retrieval-Augmented Generation (RAG) with vector search to provide answers strictly from uploaded documents.

---

Features

Spaces (Chat Workspaces)

• Create multiple independent spaces
• Rename spaces
• Delete spaces
• Track last activity per space
• Strict context isolation between spaces

Document Management

• Upload PDF documents

• Multiple documents per space

• Delete documents

• Processing status tracking:

  • processing
  • ready
  • failed

AI Chat

• Ask questions about uploaded documents
• Context-aware answers
• Conversation history support
• Source citations included in responses
• Snippet + page references for retrieved chunks

Vector Search

• Automatic document chunking
• Embedding generation
• Semantic retrieval using Qdrant
• Per-space retrieval isolation

---

Architecture

Frontend (React)
        ↓
Backend Service (Node.js + Express + PostgreSQL)
        ↓
AI Service (FastAPI)
        ↓
Qdrant Vector Database

---

Tech Stack

Frontend

• React
• TypeScript
• Vite
• LESS

Backend

• Node.js
• Express
• PostgreSQL
• TypeScript

AI Service

• FastAPI
• Sentence Transformers
• Google Gemini API / Ollama (configurable)
• LangChain-style RAG flow

Infrastructure

• Docker
• Docker Compose
• Qdrant

---

Current Capabilities

• Multi-space document analysis
• Context-aware AI chat
• PDF ingestion
• Vector search
• Source citations
• Space/document CRUD operations
• Shared storage between services
• Dockerized development environment

---

Project Structure

root/
├── backend/
├── ai-service/
├── frontend/
├── docker-compose.yml

---

Local Development

Requirements

• Docker
• Docker Compose

---

Start the Project

docker compose up --build

---

Services

Service	Port
Frontend	5173
Backend	3000
AI Service	8000
Qdrant	6333
PostgreSQL	5432

---

Environment Variables

Create .env files for backend and AI service.

Example:

GEMINI_API_KEY=
LLM_PROVIDER=gemini
GEMINI_MODEL=gemma-4-31b-it

Do not commit real secrets.

---

API Overview

Spaces

• Create space
• Get spaces
• Update space name
• Delete space

Documents

• Upload document
• Get documents for a space
• Delete document

Chat

• Ask questions
• Retrieve sources
• Persist conversation history

---

Notes

This project is currently under active development.

The focus is on building a scalable and production-oriented AI document analysis system with clean architecture and strong separation between application logic and AI processing.

---

License

MIT