Several academics have examined the various machine learning and deep learning methods with an eye on attack detection. Undoubtedly, the dataset offers extremely non-linear Network Intrusion Detection Systems (NIDS), which were developed with machine learning techniques to defend the linked IoT devices against complex botnet-based attacks. Because these organizations are always providing services to the business, the Internet of Things, which links programs, devices, data storage, and services, may open up new avenues for invasions. This research suggests a mixed deep learning method for identifying illicit software and malware-contaminated data on the Internet of Things. The argument is supported by the growing number of intrusions on Internet of Things devices and intermediary communication mediums. Long-term, undetected IoT assaults might cause major service interruptions and financial losses. It poses a threat to identity protection as well. For IoT-enabled operations to be reliable, secure, and profitable, real-time intrusion detection on IoT devices is required. Several methods have been put forth by researchers to instantly detect and identify botnets. These suggested fixes, however, struggle to keep up with the botnets’ quick evolution. In order to detect zero-day botnet attacks in real time, this study suggests a deep learning model for botnet detection.