Modern computer networks face a persistent and growing range of cyber threats, including unauthorized access, malware infiltration, denial-of-service attacks, and data exfiltration. Conventional signature- and rule-based intrusion detection systems (IDS) struggle to identify zero-day attacks and evolving intrusion patterns as networks expand in scale and complexity, and they typically require frequent manual updates to remain effective. This paper presents a machine learning–based intrusion detection system capable of automatically analyzing network traffic and distinguishing malicious activity from legitimate behavior. By learning from historical traffic data, the proposed system adapts to emerging threats with greater accuracy than static, rule-driven approaches. The system is implemented in Python using established data-analytics and machine-learning libraries, and it evaluates several supervised classifiers—Random Forest, Support Vector Machine, Naïve Bayes, and Logistic Regression—to categorize network traffic as normal or intrusive. Experimental results show that the Random Forest classifier consistently achieves the strongest overall performance, and the integration of real-time packet capture with a visualization dashboard enables continuous, interpretable monitoring. The findings demonstrate that machine learning offers a scalable, adaptive, and largely automated foundation for securing contemporary network infrastructures.