Sentinel-AI is an end-to-end security solution that captures live network traffic, processes it into machine-learning-ready features, and uses a Random Forest Classifier to detect malicious activity (such as Nmap scans) in real-time.
This project was developed to demonstrate the integration of Cybersecurity Forensics and Supervised Machine Learning.
Operating System: Kali Linux 2026 (VirtualBox environment)
Language: Python 3.10+
Libraries: * Scapy: For low-level packet sniffing and dissection.
Pandas: For data structuring and CSV engineering.
Scikit-Learn: For training the Random Forest model.
Colorama: For real-time terminal alerts.
Data Ingestion (sentinal.py): Uses Scapy to sniff raw packets and extract features like src_port, dst_port, protocol, and payload_size.
Model Training (train_ai.py): Processes the captured CSV data. It uses an 80/20 train-test split to train a Random Forest model, achieving 100% accuracy
in detecting synthetic scan patterns.
Live Defense (live_defender.py): A real-time monitoring script that applies the trained model to live traffic, triggering visual alerts upon detecting anomalies.
Detection Accuracy: 100% (Lab Environment)
Detected Threats: TCP Connect Scans, Port Discovery, and Anomalous Payload sizes.
Alert System: Integrated color-coded terminal outputs for immediate incident response.
Name: [Manish Kumar]
Background: M.Sc. IT | Certified Ethical Hacker (CEH) | Specialized in GovTech & AI Security.