The increasing frequency of advanced persistent threats (APTs) in the dynamic sector of cybersecurity calls for new techniques to increase network security. This work constructs and evaluates a sophisticated network traffic monitoring system that uses a Kernel Function to train a nonlinear support vector machine (SVM) in order to counteract APTs. Current models often fail to provide comprehensive and accurate threat detection. Recall of 94%, accuracy of 92%, precision of 88%, and F1-score of 0.95 show how much better our proposed system performs. Scalability and quick processing of large-scale datasets are ensured by the connection with Apache Spark MLlib, which is a crucial part of real-time threat detection. This work presents a novel integration of nonlinear support vector machines (SVMs) with distributed computing, resulting in a robust and adaptive method that outperforms earlier models in APT detection with respect to accuracy and reliability. The measurements that have been provided and displayed in a thorough bar chart help our suggested model stand out from the competitors. This study not only creates a new technique for detecting APTs but also sets a new benchmark for network security efficacy. This will enable more resilient cybersecurity frameworks to survive evolving threats.