Here’s your updated README.md with architecture included and clean structure (single copy-ready format):

---

Vecta – Agentic Multi-PDF RAG System (Next.js + Gemini + Endee)

A production-ready full-stack Retrieval-Augmented Generation (RAG) application that enables users to upload multiple PDFs, store embeddings in a vector database, and perform intelligent agentic reasoning using Google Gemini AI through a modern, interactive UI.

---

🚀 Tech Stack

• Frontend: Next.js 14 (App Router), Tailwind CSS, Framer Motion, React Markdown
• Backend: FastAPI, google-generativeai, pymupdf, sentence-transformers
• Database: Endee Vector Database (Docker)

---

🏗️ Architecture

                ┌──────────────────────────────┐
                │          Frontend            │
                │  Next.js + Tailwind + FM     │
                │  (Chat UI + Upload UI)       │
                └──────────────┬───────────────┘
                               │
                               │ API Calls (/api/*)
                               ▼
                ┌──────────────────────────────┐
                │        FastAPI Backend       │
                │                              │
                │  • PDF Processing (PyMuPDF)  │
                │  • Chunking                  │
                │  • Embeddings (MiniLM)       │
                │  • Query Orchestration       │
                └──────────────┬───────────────┘
                               │
             ┌─────────────────┴─────────────────┐
             │                                   │
             ▼                                   ▼
┌──────────────────────────┐        ┌──────────────────────────┐
│   Endee Vector DB        │        │     Gemini 1.5 Pro AI     │
│  (Embeddings Storage)    │        │  (Reasoning + Response)   │
└──────────────────────────┘        └──────────────────────────┘
             │
             ▼
   Retrieved Context Chunks
             │
             ▼
     Final Answer + Sources
             │
             ▼
         Frontend UI

---

⚙️ Setup Instructions

🔑 Environment Variables

Backend (backend/.env)

GEMINI_API_KEY="your_google_ai_studio_api_key_here"
ENDEE_HOST="localhost"
ENDEE_PORT="8080"

---

🐳 Endee Setup (Docker)

docker run \
  --ulimit nofile=100000:100000 \
  -p 8080:8080 \
  -v ./endee-data:/data \
  --name endee-server \
  --restart unless-stopped \
  endeeio/endee-server:latest

---

⚡ Backend Initialization

cd backend
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt
uvicorn main:app --reload --port 8000

Backend runs on: 👉 http://localhost:8000

---

💻 Frontend Initialization

cd frontend
npm install
npm run dev

Frontend runs on: 👉 http://localhost:3000

Note: /api/* routes are proxied to FastAPI via Next.js rewrites.

---

🔌 API Routes

📂 POST /api/upload

• Accepts: multipart/form-data with one or more PDF files

• Process:

  • Extract text using PyMuPDF
  • Chunk content
  • Generate embeddings (MiniLM-L6-v2)
  • Store in Endee vector database

---

💬 POST /api/query

• Accepts:

{
  "query": "your question",
  "history": []
}

• Returns:

  • AI-generated answer (Gemini)
  • Source metadata (file name + page references)

---

🧠 Core Workflow

1. User uploads PDFs
2. Backend extracts & chunks text
3. Embeddings are generated and stored in Endee
4. User asks a query
5. Relevant chunks are retrieved from Endee
6. Context + query sent to Gemini
7. Gemini generates grounded response
8. Answer + sources displayed in UI

---

✨ Features

• Multi-PDF upload and semantic search
• Agentic reasoning with Gemini AI
• Context-aware conversational memory
• Source citation for transparency
• Smooth animated UI (Framer Motion)
• Fully local vector database (Endee)

---