AI-Powered Logistics Fleet Optimization System
DeepFleet-AI is an end-to-end, production-grade fleet management and route optimization platform built using Machine Learning, AWS Cloud, and React. It enables intelligent delivery routing, ETA prediction, and real-time fleet visibility using historical delivery data and live tracking.
- Python, Flask
- Scikit-learn, Pandas, NumPy, XGBoost
- Geopy, Haversine, Folium
- React.js, Tailwind CSS, Recharts
- AWS EC2, S3, CloudWatch, Lambda
- Docker, GitHub Actions
DeepFleet-AI/
├── backend/
│ ├── main.py # FastAPI entrypoint
│ ├── routes/
│ │ ├── fleet.py
│ │ └── auth.py
│ ├── services/
│ │ └── optimizer.py # Route optimization logic
│ ├── models/
│ │ └── delivery_model.py # SQLAlchemy models
│ └── utils/
│ └── geo_utils.py # Geolocation, Haversine formula
│
├── ml/
│ ├── model.py # Fleet routing ML model
│ ├── train.py # Training script
│ ├── predict.py # Inference logic
│ └── data/
│ └── delivery_logs.csv # Sample training data
│
├── frontend/ # Optional React Dashboard
│ └── src/
│ └── components/
│ └── pages/
│ └── Dashboard.jsx
│
├── docker/
│ └── Dockerfile
│ └── docker-compose.yml
│
├── scripts/
│ └── seed_db.py # Populate DB
│ └── scheduler.py # Fleet update scheduler (Lambda)
│
├── infrastructure/
│ └── terraform/ # AWS setup
│ └── sagemaker-deploy.tf
├── api/ # Flask backend APIs
│ ├── app.py # API entrypoint
│ ├── routes/
│ │ └── delivery_routes.py
│ └── utils/
│ └── helpers.py
├── README.md
└── requirements.txt - ✅ Predict ETA (Estimated Time of Arrival) using ML
- ✅ Route Optimization using Haversine/Mapbox APIs
- ✅ Vehicle Type Classification
- ✅ Data ingestion from CSV or REST
- ✅ Scalable deployment with Docker
git clone https://github.com/hq969/DeepFleet-AI.git
cd DeepFleet-AIcd api/
python -m venv venv
source venv/bin/activate
pip install -r ../requirements.txt
python app.pycd frontend/
npm install
npm run devdocker build -t deepfleet-backend .
docker run -p 5000:5000 deepfleet-backendEndpoint: /predict_eta
POST Body:
{
"origin": [28.6139, 77.2090],
"destination": [28.5355, 77.3910],
"vehicle_type": "van"
}Response:
{
"eta_minutes": 42.5
}Sample delivery logs can be found in:
ml/data/delivery_logs.csv
You can expand this with more real-world logs.
- Integrate Mapbox/Google Directions API
- Add MongoDB or PostgreSQL for delivery log storage
- Real-time GPS data streaming via AWS Kinesis
- Harsh Sonkar — LinkedIn
MIT License. Free to use and modify with attribution.