Android became one of the world's most widely used mobile platforms, with an overall market share of 71.45%. As a result, malicious software developers are increasingly concentrating their efforts on Android-based devices. The amount of malicious software that targets mobile devices is growing exponentially each year. Android lets users download and install programs for their operating system from unverified sources. As a result, it is now quite simple for those who create malware to bundle it together with legitimate apps and then spread it. Traditional malware detection approaches frequently fail to identify harmful applications. As a result, Android malware detection needs a more dependable, accurate, generalizable, smoothly deployable, and efficient model. We develop a lightweight, easily deployable, efficient behavior-based Android malware detection system. We set up our emulator to prevent it from getting detected by the emulator-aware malware. We use a random input generation-based monkey tool with 4000 events to maximize the code coverage. We extract permission, intents, and system calls from an APK file to determine whether an application is benign or malware. We use three feature selection techniques applied sequentially to reduce the extensive feature set and select only highly informative features. We achieve an accuracy of 97% and an F1-score of 97.04% using the XGBoost classifier. Finally, we test our model on entirely new obfuscated malware samples and achieve an accuracy of 81%.