AI-Driven Infrastructure Protection Framework for Resilient Enterprise Networks

This paper presents an AI-driven infrastructure protection framework to enhance the resilience of enterprise networks. It integrates machine learning, threat intelligence, and cloud-native orchestration to detect threats, profile behaviors, and automate responses. The architecture ingests network logs and telemetry, applies anomaly detection and risk scoring, and correlates results with threat intelligence for real-time policy enforcement. Evaluation using CICIDS 2017 & 2020 datasets shows the framework outperforms traditional intrusion detection systems in accuracy and responsiveness. LSTM and Random Forest models achieved the best results, confirmed through ROC and confusion matrix analysis. Feature importance insights and a dynamic risk scoring engine support scalable and context-aware decision-making. This work demonstrates the effectiveness of combining AI with cloud-native defense for proactive, intelligent cybersecurity. Future extensions will explore explainable AI, federated learning, and adversarial robustness.