DDoS attacks have become an important concern in the security of networks because of the growing dependenceondigital communication systems and cloud-based services. The paper gives the design and implementation of a smart DDoSattackdetection framework using machine learning techniques. The major aim is to establish a system that will be abletoclassifynetwork traffic data accurately to either normal or malicious traffic. The suggested framework also includes some critical stepssuch as preprocessing of data, extraction of features, and training of the model, which is necessary to achieve better detection.Several machine learning techniques, e.g. Support Vector Machine, Random Forest, and K-Nearest Neighbors areusedtoconstruct classification models. These models are tested and contrasted on the basis of the conventional performancemeasuressuch as accuracy, precision, recall and F1-score to establish the most effective way to detect DDoS. Through experimental findings,it is proved that ensemble-based and tree-based classifier are better than traditional classifiers in detection of complexattackpatterns at lower false positive rates. The system will be a scalable and efficient system, which will facilitate real-time detectioninan active network environment. The proposed framework offers beneficial information on the choice of the best models tobeusedin intrusion detection systems by making use of comparative analysis of alternative algorithms. The study helps inintensifyingthesecurity measures against cybercrimes by providing a trusted and intelligent mechanism to counter early detectionandcountermeasures of DDoS attacks in contemporary network systems. Keywords: DDoS Attack Detection, Machine Learning, Intrusion Detection System, Network Security, Feature Selection, Real-Time Detection.