Brain computer interface is employed in several applications providing explicit mode of communication between the brain and computers. Particularly, electroencephalography (EEG) is one of the most conventional methods for acquiring visual evoked potentials that is acquired from external stimuli, such as the P300 speller elicits the P300 potential from the presentation of characters and symbols. By employing machine learning classifiers and P300 potential has shown promising results for identifying and authenticating users since the brainwaves generated by each person while facing a particular stimulus are distinctive. But the current authentication research have not fully explore the P300 potentials and are not very successful when analyzing the most suitable processing and machine learning based classification techniques. In this study an approach for user recognition scheme utilizing the P300 speller is proposed to validate it on 8 users creating a non-invasive EEG based user authentication scheme. This framework achieved a performance of 100% accuracy in user recognition for the deep neural network (DNN) classifier, highlighting its effectiveness in accurately identifying and authenticating users thus, indicating the probability of performing EEG based user authentication using P300 speller paradigm.