Epilepsy is a neurological disorder that affects millions of people worldwide, especially in low-resource settings where access to specialized medical care is limited. This study proposes a model for classifying epileptic patterns in EEG signals using advanced deep learning techniques. The methodology includes phases such as data preparation, signal preprocessing to reduce noise, feature extraction using time-frequency domain analysis, and model training with a hybrid architecture that combines convolutional neural networks (CNNs) and transformer-based models to improve pattern recognition. Extensive experimentation with various deep learning architectures and hyperparameter configurations culminated in the identification of a highly effective model for epilepsy detection, a CNN composed of three one-dimensional convolutional layers followed by a flattening layer. This model demonstrated exceptional performance, achieving an accuracy of 0.9983 and an AUC-ROC of 0.99995, underscoring its near-perfect discriminatory ability to distinguish epileptic patterns from non-epileptic EEG signals. In conclusion, the model’s remarkable performance indicates technical excellence and clinical reliability, as the model consistently identifies seizure-related anomalies with minimal false positives or negatives.