Facial recognition is a critical task in computer vision with applications in various domains such as security, biometrics, and human-computer interaction. Our research explores the applications of reinforcement learning (RL) techniques and pre-trained models for face recognition tasks. The objective was to develop a robust facial recognition system without extensive training on custom data. Experiments were conducted on dataset consisting of many images, using the DeepFace deep learning (DL) model in combination with the DLib computer vision library. We fine-tuned the pre-trained model parameters and optimized its performance using RL. The proposed methodology was able to an accuracy of 97.81%. Our results highlight the effectiveness of RL in adapting pre-trained models for face recognition tasks. The combined methodology offers a promising approach for rapid deployment and scalability of face recognition systems. Future research directions may include further refinement of the methodology, exploring different architectures, and addressing privacy and ethical considerations in face recognition systems.