AuraPilot (Javis-Bot)

🎯 AuraPilot is an AI-powered document intelligence assistant — upload documents, index them, and ask natural-language questions that are answered using Retrieval-Augmented Generation (RAG).

---

🚀 Quick Overview

• Backend: FastAPI (Python) serving REST endpoints and document processing, embeddings, and RAG flow.
• Frontend: Vite + React + shadcn/ui (TypeScript) providing a modern UI for login, document upload, and chat.
• Vector DB: Pinecone (serverless indices) for storing embeddings and performing semantic search.
• Database: Supabase for user and document metadata.
• Embeddings: sentence-transformers (model configurable via ENV); fallback to mock embeddings for local dev.
• LLM: Ollama (local LLM) used for generating responses using RAG contexts.

---

🧭 Architecture & Workflow

1. User uploads a document via the frontend.
2. Backend (DocumentProcessor) extracts and chunks text from the document.
3. Each chunk is embedded via embedding_service.
4. Vectors + metadata are upserted into Pinecone, partitioned by user namespace (e.g., user_1).
5. Document metadata is stored in Supabase for listing and tracking status.
6. When a user asks a query:
   • Query embedding is generated
   • Pinecone is queried for the top-K similar chunks
   • Chunks are concatenated into a context (bounded by MAX_CONTEXT_LENGTH)
   • The context and query are sent to the LLM (Ollama) to generate the final answer

---

🛠️ Getting Started (Local Development)

Prerequisites

• Node.js (recommended), npm or bun
• Python 3.11+
• A Pinecone account (API key) — optional for local dev
• Supabase project (optional for metadata persistence)
• Ollama running locally for LLM (or configure alternative)

Backend

1. Create a virtual env and activate it

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

2. Copy .env.example (or create .env) and set environment variables (see ENV section below).
3. Start the backend:

python -m uvicorn app.main:app --host 0.0.0.0 --port 8001

Frontend

cd verdant-ai-chat
npm install
# or use bun: bun install
npx vite
# open http://localhost:8080

---

⚙️ Environment Variables (important ones)

• SUPABASE_URL, SUPABASE_KEY, SUPABASE_SERVICE_ROLE_KEY
• PINECONE_API_KEY, PINECONE_ENVIRONMENT, PINECONE_INDEX_NAME, PINECONE_INDEX_DIMENSION
• OLLAMA_API_URL, OLLAMA_MODEL
• EMBEDDING_MODEL (default all-MiniLM-L6-v2)

These are defined in backend/.env and loaded by pydantic-settings.

---

🔎 Useful Endpoints

• Document upload: POST /api/v1/documents/upload
• List documents: GET /api/v1/documents?user_id=1
• Pinecone health/status: GET /api/v1/documents/health/pinecone-status

---