A simple FastAPI-based threat detection API that uses a trained model to predict if network traffic is malicious based on the number of bytes transferred.

Suresh

This project is a minimal implementation of a machine learning model served using FastAPI. It accepts a value for bytes_transferred and returns a prediction with a confidence score on whether the network traffic is malicious.

• FastAPI: for building the API
• Joblib: for model loading
• Scikit-learn: for the ML model (RandomForest)
• Uvicorn: ASGI server to run FastAPI

.
├── app
│   ├── main.py         # Main FastAPI app
│   └── model.pkl       # Trained model file
├── requirements.txt    # Python dependencies
├── README.md
└── .gitignore

• Method: POST
• Request Body:

{
  "bytes_transferred": 4500
}

• Response:

{
  "verdict": "benign",
  "confidence": 0.87
}

1. Clone the repo:

git clone https://github.com/suri10/ai-threat-detector.git
cd ai-threat-detector

2. Create a virtual environment:

python -m venv venv
source venv/bin/activate  # or .\venv\Scripts\activate on Windows

3. Install dependencies:

pip install -r requirements.txt

4. Run the app:

uvicorn app.main:app --reload

• Build and test prediction model
• Create FastAPI app
• Connect /predict endpoint
• Set up for GitHub upload
• Deploy to cloud (Future)

Pull requests are welcome!

This project is open source and available under the MIT License.

Let's catch threats with AI! 💻⚡