Attention Deficit Hyperactivity Disorder (ADHD) represents a widely prevalent neurodivergence. Current diagnostic approaches rely on subjective symptom assessment, leaving room for improvement through objective, biology-informed decision support. EEG-based machine learning classifiers have been proposed to distinguish ADHD and neurotypical individuals, but results are inconsistent, and applicability in a clinical setting remains unclear. A CNN model with temporal and spatial filtering using EEG recordings to classify ADHD individuals was developed, outperforming the SVM baseline model. Interpretability techniques revealed that the CNN model learned meaningful features in line with neurophysiological ADHD studies, with frequency features being the most informative. This work demonstrates the proof of concept for objective EEG-based ADHD classification using a CNN model.