The widespread use of the internet has led to various cyber threats that can impact everyday life. These threats pose serious risks to privacy, security, and economic stability. As cyberattacks become more advanced and frequent, there is an increasing need for improved systems that can quickly and effectively detect these threats. In this work, we present a new machine-learning framework designed to detect malicious internet activity in real-time. An intrusion detection system that combines machine learning techniques with network forensics is introduced to provide a robust cybersecurity solution. Using a large network traffic dataset and employing techniques to balance uneven data, our system improves the accuracy of identifying various cyber threats, including DDoS attacks, port scans, and infiltrations.