The Internet of Things has enabled intelligent automation and seamless connectivity, revolutionizing several industries. But this widespread adoption has also made botnet attacks more likely, which poses serious risks to IoT security. Owing to the distinct features and limitations of IoT environments, conventional security protocols frequently need to advance. This research proposes a novel framework to enhance botnet attack detection in Internet of Things systems through the integration of deep learning and ensemble learning techniques. Our methodology, which makes use of the N-BaIoT dataset, includes extensive preprocessing of the data, feature extraction, and the use of sophisticated classification techniques. The architectural design of the framework includes gathering data from network botnet attacks, debugging, extracting features, preprocessing, and using ensemble models and deep learning for feature selection and classification. Superior detection accuracy and robustness were demonstrated by integrating models such as Convolutional Neural Networks, Recurrent Neural Networks, Random Forest, and Gradient Boosting. Our findings show 98.7% detection accuracy, which is superior to methods using a hybrid approach. This study emphasizes how deep learning and ensemble learning approaches can be combined to strengthen IoT systems’ defenses against botnet attacks. The suggested approach offers a reliable and scalable means of identifying and reducing botnet attacks, which constitutes a noteworthy development in IoT security.