Malicious applications targeting mobile devices are very prevalent today. In addition, the open-source nature of the Android platform and many unofficial app stores have increased its attractiveness and allowed for the proliferation of malware because it can easily be downloaded from such unofficial stores. To complement this aspect, the research introduces ML-PAAMI, which will constitute a novel permissions-based malware detection system that can identify potentially harmful applications by analyzing suspicious permissions. Our layered methodology emphasizes key features, including identifying high-risk and safe permissions and permission ratings directly obtained from Android Application Package (APK) files through Application Programming Interface (API) requests. We employ machine learning models—such as Neural Networks, Support Vector Machines, and Logistic Regression—to classify applications as malicious or benign. After explicating experiments and tests, ML-PAAMI obtained a high accuracy rate of 93.69% by utilizing the Support Vector Classifier and outperforming other algorithms comparatively. The system also optimizes feature selection, leading to good metrics such as precision, accuracy, recall, and F-measure. ML-PAAMI is an open-source solution that presents a helpful tool for Android malware detection.