The study presents the potential of using a simplified EEG-based Brain-Computer Interface (BCI) for subject recognition by developing and comparing two classification models built on the custom “Login” dataset. The first is a binary per-subject model utilizing the Fast Tree classifier, while the second is a multiclass model based on LightGBM. Both classifiers were fine-tuned by optimizing key hyperparameters, such as the number of trees, learning rate, and number of iterations. The preprocessing steps included a poor signal level filter and the Simple Moving Average algorithm, both of which were evaluated to determine optimal parameters to improve classification performance. Feature importance analysis revealed that the EEG bands, particularly alpha, beta, and gamma, had the most significant impact, with gamma showing a strong influence in the multiclass model. The evaluation results demonstrated high performance: the binary classifier achieved an F1 score of 0.927 and an accuracy of 0.985, while the multiclass model reached macro and microaccuracies of approximately 0.929. The findings suggest that subject-specific binary models may offer advantages in practical applications, especially in scalable or commercial settings where the ease of onboarding new users is critical.