🌿 TenderGreen AI — Tender Sustainability Analyzer

AI-Powered analysis of government tenders for SDG 11 & SDG 12 compliance

Upload a government tender PDF → AI extracts environmental clauses, evaluates sustainability commitments, and identifies missing SDG requirements with a visual compliance scorecard.

🎯 SDG Focus

• SDG 11 — Sustainable Cities and Communities
• SDG 12 — Responsible Consumption and Production

🏗️ Architecture

Frontend (Next.js) → FastAPI Backend → PDF Processor → RAG Pipeline → Gemini LLM
                                        ↑                  ↑
                                    PyMuPDF/OCR      ChromaDB + HuggingFace

✨ Features

Feature	Description
📄 Smart PDF Extraction	Hybrid OCR + native text extraction
🤖 RAG Analysis	Grounded analysis using actual tender text
🎯 SDG Gap Detection	Identifies missing sustainability requirements
📊 Visual Scorecard	Interactive radar charts for SDG coverage
💬 Chat with Tender	Ask questions about your tender document
📋 Recommendations	Actionable clause suggestions for gaps

🚀 Quick Start

Prerequisites

• Python 3.10+ (conda env: rl_env)
• Node.js 18+
• Google Gemini API key (free from Google AI Studio)

1. Backend Setup

cd backend

# Set your API key
# Edit .env and replace 'your_gemini_api_key_here' with your actual key
notepad .env

# Install dependencies (if not already done)
conda activate rl_env
pip install -r requirements.txt

# Start the server
python -m uvicorn app.main:app --host 0.0.0.0 --port 8000 --reload

2. Frontend Setup

cd frontend

# Install dependencies (if not already done)
npm install

# Start dev server
npm run dev

3. Open the App

• Frontend: http://localhost:3000
• Backend API: http://localhost:8000
• API Docs: http://localhost:8000/docs

📂 Project Structure

internship_2/
├── backend/
│   ├── app/
│   │   ├── main.py              # FastAPI entry point
│   │   ├── config.py            # Environment config
│   │   ├── models/schemas.py    # Pydantic models
│   │   ├── services/
│   │   │   ├── pdf_processor.py # PDF extraction engine
│   │   │   ├── rag_pipeline.py  # RAG + ChromaDB
│   │   │   └── sdg_analyzer.py  # Gemini LLM analysis
│   │   └── knowledge/           # SDG knowledge base JSONs
│   ├── sample_tenders/          # Sample tender PDFs
│   ├── .env                     # API keys (gitignored)
│   └── requirements.txt
├── frontend/
│   ├── src/app/
│   │   ├── page.tsx             # Landing page
│   │   ├── analyze/page.tsx     # Upload + Dashboard
│   │   └── globals.css          # Design system
│   └── src/lib/
│       ├── api.ts               # Backend API client
│       └── types.ts             # TypeScript interfaces
└── README.md

🧪 Sample Tenders

Two sample tenders are included for testing:

1. Highway Construction — MORTH NH-48 (moderate sustainability)
2. Smart City ICCC — Mumbai Smart City (better sustainability)

Generate them:

cd backend
conda activate rl_env
python generate_sample_tenders.py

📡 API Endpoints

Method	Endpoint	Description
POST	/api/upload	Upload tender PDF
GET	/api/status/{id}	Check processing status
GET	/api/analysis/{id}	Get full analysis results
POST	/api/analysis/{id}/chat	Chat with tender
GET	/api/health	Health check

🔧 Tech Stack

• Frontend: Next.js 16, React 19, TypeScript, Recharts, Vanilla CSS
• Backend: FastAPI, Python 3.10
• AI: Google Gemini 2.0 Flash, LangChain, ChromaDB, HuggingFace sentence-transformers
• PDF: PyMuPDF, pdfplumber, Tesseract OCR (optional)