In an era of escalating cyber threats, having effective network security is vital for preserving sensitive data and digital assets. In order to improve threat identification and mitigation, this study suggests a robust network security framework that uses a dual- layered, hybrid model combined with machine learning. The framework uses machine learning algorithms to continuously adjust to changing attack patterns by combining signature-based and anomaly- based intrusion detection techniques. By combining deep packet inspection and perimeter defense mechanisms, the dual-layered strategy improves security and provides complete defense against known and undiscovered threats. Additionally, the hybrid approach reduces false positives and increases threat classification accuracy by combining sophisticated AI-driven analytics with rule-based heuristics. Results from experiments show how well this framework works to identify cyberthreats with high accuracy Through the presentation of an intelligent, flexible, and robust security architecture, this study advances network security.