The number of vehicles is increasing rapidly due to the advancements in modern transportation systems. However, this growth has led to several issues that disrupt smooth and safe traffic flow on roads, such as casual driving, over-speeding, frequent lane changes, and illegal overtaking. Fatigue due to overworking and poor driving behavior are among the leading causes of accidents. Therefore, driver self-awareness plays a key role in reducing accidents. This paper proposes a novel approach to driver behavior analysis and profiling by analyzing data from cameras and inertial sensors. The system includes a camera to extract facial landmark points to detect signs of drowsiness and other symptoms of fatigue. Moreover, the system uses the MPU-9255 sensor to collect accelerometer, gyroscopic, and magnetometer data attached to Raspberry Pi to identify driving maneuvers such as aggressive braking and cornering. The system uses Long Short-Term Memory (LSTM) based architecture, which effectively learns patterns from the sensor data. By analyzing these events, the system can give an alert to the drivers and prompt them to take necessary actions. These measures can reduce the potential risk of accidents. The results of the proposed methodology, tested on the Driver Behavior Dataset and collected data using the system, indicate robustness and efficiency.