Project Review and Governance AI-based Transparency Interface
An enterprise-grade AI platform for automated Detailed Project Report (DPR) analysis with ML-driven feasibility assessment and governance compliance validation.
| Component | Stack | Purpose |
|---|---|---|
| ML Engine | XGBoost 3.0.5 + scikit-learn 1.7.2 | Feasibility classification (91% accuracy) |
| NLP Pipeline | TF-IDF + pandas 2.3.0 | Feature extraction & text vectorization |
| Document Processing | pdfminer.six + pytesseract | PDF parsing & OCR processing |
| Backend API | Flask 2.3.3 | RESTful services & CORS handling |
| Frontend | React 18.2.0 + Chart.js 4.4.0 | Interactive dashboard with real-time viz |
| Charts Integration | react-chartjs-2 3.3.0 | Doughnut, Bar, Line charts for insights |
XGBClassifier(
n_estimators=300, # Ensemble size
learning_rate=0.1, # Gradient step size
max_depth=6, # Tree complexity
tree_method="gpu_hist", # GPU acceleration
predictor="gpu_predictor"
)- Text Features: TF-IDF vectors (5K features, 1-3 grams)
- Numerical Features: Budget analysis, timeline estimation
- Domain Features: MDONER compliance score, technical feasibility
- Output: Probability distributions + confidence scoring (0-100%)
- Size: 110 DPR entries with ground truth labels
- Classes: Feasible (60%) vs Risky (40%) projects
- Domains: Infrastructure, Healthcare, Education, Tourism, Technology
- Geography: Northeast India (rural + urban projects)
| Component | Minimum | Recommended |
|---|---|---|
| CPU | 2 cores, 2.4 GHz | 4+ cores, 3.0 GHz |
| RAM | 8 GB | 16 GB |
| Storage | 5 GB available space | 10 GB SSD |
| GPU | Not required | NVIDIA CUDA-compatible |
| Software | Version | Required For |
|---|---|---|
| Python | 3.11+ | Backend, AI, Text Processing |
| Node.js | 18+ | Frontend development |
| Git | Latest | Version control |
| Visual C++ Build Tools | Latest | Python package compilation |
- CI/CD: GitHub Actions with auto-approval (75% success threshold)
- Monitoring: Comprehensive logging with performance metrics
- Security: CORS configuration, input validation, sanitization
- Scaling: Multi-worker deployment ready
Built for Smart India Hackathon 2024 | Transforming governance through AI