Android botnets have become a significant threat due to the platform’s open-source nature and widespread usage. This study describes the deployment of an Android botnet detection system based on machine learning, employing a hybrid methodology that enhances detection accuracy by combining static and dynamic analysis. Static analysis identifies malicious behaviors through permissions and API calls, while dynamic analysis monitors network traffic and system calls. Using performance criteria like accuracy and F1 score, the system assesses the Random Forest, XGBoost, and SVM algorithms and chooses the top-performing model. The maximum detection accuracy for botnets in our solution is 99.4%.To improve the system’s performance in real-time detection, important issues such feature selection and data imbalance were resolved. The goal of future research will be to increase flexibility to changing malware threats.