This paper proposes a methodology for age-invariant face recognition (AIFR) using optimized deep learning features, specifically through transfer deep learning and a genetic algorithm for feature selection. The system utilizes the VGG-Face model to extract compact face features without preprocessing, achieving significant recognition rates of 86.2% and 96% on FGNET and Morph datasets respectively. The combination of genetic algorithm optimization and K-nearest neighbor classification demonstrates notable improvements over existing methods.