Cybersecurity

NetMarshal – AI Network Automation

NetMarshal is an autonomous network management framework designed to replace manual network administration tasks. It continuously monitors network traffic, classifies devices and applications using machine learning, and automatically pushes QoS or security policies to routers and switches.

Project typeIndependent build
StatusCompleted project
FocusCybersecurity

Technology

NFStreamScikit-learnNetmikoStreamlitGNS3Python

System approach

  • Traffic Sensing: NFStream for deep packet inspection and flow extraction.
  • Intelligence Engine: Scikit-learn Random Forest models for traffic classification.
  • Actuation: Netmiko for pushing CLI commands to network devices.
  • Dashboard: Streamlit for real-time visualization.

Challenges

  • Normalizing traffic features across different network environments.
  • Safely pushing configuration changes to live network hardware without causing downtime.
  • Achieving high accuracy in encrypted traffic classification.

Solutions

  • Standardized flow features (packet sizes, inter-arrival times) independent of IP/Port.
  • Implemented a dry-run feature and a rollback mechanism for all Netmiko automated configurations.
  • Focused on statistical flow analysis (behavioral patterns) rather than deep packet inspection, allowing classification of encrypted SSL/TLS traffic.

Key outcomes

01

Autonomous network control

02

ML-based device classification

03

Real-time QoS optimization