The loss of an arm significantly impacts a person’s independence and a struggle to perform the daily activities. Bionic prosthetic arm has been life-changing for these people. The introduction to various control strategies such as Electromyography (EMG) has transformed the bionic arm by making it functional, responsive and precise. This thesis mainly focuses on the EMG as control strategy which helps people to control the myoelectric arm with muscle intention just like a normal arm. The EMG signals from the sensors, SHIMMER EMG and Arduino Myoware, are captured using software MATLAB and Arduino IDE. These signals are later processed and used to control the gesture of the arm using the same software. The study focused on EMG signals from the forearm muscles such as Extensor digitorum and Flexor carpi radialis for training the model using the MATLAB Toolbox. The training includes simple hand gestures such as hand open, hand close and flexion of all five fingers individually. Feature Extraction performed for time-domain features such as the Root Mean Square (RMS), Mean Absolute Value (MAV) and Waveform Length (WL) for the signal. Random Forest classification algorithm is implemented to classify the hand gestures achieved approximately 80% accuracy. The classified hand gestures for open and close hand were tested out on a prosthetic finger using Arduino Uno microcontroller to control the motor. The feature extraction, training, classification and serial communication which the Arduino microcontroller was all performed on MATLAB. The model is being trained more using more dataset to increase the accuracy for better performance.