Multi-biometric systems have become increasingly popular in biometrics as they improve security and accuracy in identity verification. The research introduces a new method for a multi-biometric system that combines fingerprint, iris, and face biometric modalities. This approach uses a hybrid deep learning classifier and optimization algorithm that is specifically designed for this purpose. The goal is to develop a system that is reliable and efficient by combining the advantages of different modalities. This will enhance the accuracy of recognition and decrease the risk of spoof attacks.