SahayakAI — Multilingual AI Knowledge Vault

Your Neural Knowledge Vault · Hindi + English · RAG Chatbot with FAQ Cache & Smart Tools

Stack: FastAPI · Gemini Flash · ChromaDB · Python 3.11+ Deploy: Render (free-tier friendly with 4–5 Gemini API keys)

---

✨ Key Features

• 💬 Multilingual RAG Chat — answers from your documents in Hindi & English
• ⚡ FAQ Token-Saver Cache — similar questions answered instantly, zero Gemini tokens
• 🧠 Query Intelligence — sentiment, intent detection & harmful-query blocking
• 🛠️ Smart Tools — document summarizer, quiz generator, translator, analytics
• 🌐 General + RAG modes — chat works even with no documents uploaded
• 📄 Multi-format — PDF, DOCX, TXT, MD, CSV
• 🔑 Multi-key rotation — spreads load across 4–5 free Gemini keys

---

📁 Project Structure

NeuraVault/
├── backend/
│   ├── main.py                       # FastAPI entry point
│   ├── config.py                     # Settings + API key pool
│   ├── requirements.txt
│   ├── .env.example
│   ├── routers/
│   │   ├── chat.py                   # POST /api/chat
│   │   ├── documents.py              # Upload / list / delete
│   │   ├── tools.py                  # Summarize / quiz / translate / FAQ / analytics
│   │   └── health.py                 # GET /api/health
│   └── services/
│       ├── gemini_service.py         # Key rotation + chat + embeddings
│       ├── chroma_service.py         # Vector store CRUD
│       ├── ingest_service.py         # Parse → chunk → embed → store
│       ├── rag_service.py            # Retrieve → prompt → answer (+ tools)
│       ├── faq_service.py            # Semantic FAQ cache (token saver)
│       └── intelligence_service.py   # Sentiment / intent / harm detection
├── frontend/
│   └── index.html                    # Chat + Smart Tools UI (served by FastAPI)
├── render.yaml                       # Render deploy config
└── WHATS_NEW.md                      # Feature changelog + SIH demo script

---

🚀 Local Setup

cd backend
python -m venv venv
source venv/bin/activate          # Windows: venv\Scripts\activate
pip install -r requirements.txt

cp .env.example .env
# Add your Gemini keys to .env (get free keys: https://aistudio.google.com/app/apikey)

uvicorn main:app --reload --host 0.0.0.0 --port 8000

Open http://localhost:8000

---

☁️ Deploy on Render

1. Push this folder to GitHub
2. Render → New → Web Service → connect repo
   • Root Directory: backend
   • Build: pip install -r requirements.txt
   • Start: uvicorn main:app --host 0.0.0.0 --port $PORT
3. Add environment variables (GEMINI_API_KEY_1…_5, CHROMA_PERSIST_DIR=/data/chroma_db)
4. Add a Disk → mount /data (1GB) so ChromaDB + FAQ cache persist
5. Deploy → live at https://neuravault-xxxx.onrender.com

---

🌐 API Endpoints

Method	Endpoint	Description
GET	/api/health	Health + key count
POST	/api/chat	Ask a question (RAG + cache)
POST	/api/documents/upload	Upload a document
GET	/api/documents	List documents
DELETE	/api/documents/{file}	Delete a document
POST	/api/tools/summarize	Summarize a document
POST	/api/tools/quiz	Generate quiz
POST	/api/tools/translate	Translate text
POST	/api/tools/followups	Suggest follow-up questions
GET	/api/tools/analytics	Vault analytics
GET	/api/tools/faqs	Popular cached questions
GET	/api/tools/faq-stats	Cache stats (tokens saved)

---