Facial age group classification is a critical task in computer vision with broad applications in healthcare, security, and demographic analysis. However, achieving accurate and fair classification remains challenging, particularly for underrepresented populations such as individuals with Black facial features. This study addresses these limitations by developing a hybrid optimisation framework that integrates Differential Evolution (DE) and Reinforcement Learning (RL) to fine-tune a Convolutional Neural Network (CNN) model for facial age classification. The developed DE-RL approach leverages DE’s global search capabilities and RL’s adaptive learning to automatically discover optimal hyperparameter configurations, including learning rate, dropout rate, number of unfrozen layers, and batch size. The system was trained and evaluated using racially inclusive facial datasets (UTKFace and CASIA-Face-Africa), ensuring demographic diversity and balanced age group representation. Experimental results demonstrate that the DE-RL optimised CNN achieved a classification accuracy of 94.4%, with precision, recall, and F1-scores above 0.94 across nine age categories. The mean absolute error was reduced to 0.16 label units, indicating improved generalisation and robustness. This research contributes to the development of equitable and high-performing facial analysis systems by addressing dataset imbalance and optimisation challenges. The DE-RL strategy proves effective in enhancing classification accuracy, fairness, and adaptability, paving the way for more responsible deployment of AI in age-sensitive applications.