We present a novel quantum-inspired deep learning model for face recognition that draws key ideas from quantum interference yet operates entirely on standard GPUs. By introducing complex-valued attention mechanisms within a lightweight convolutional neural network (CNN), our approach emulates wave-like interference effects using sinusoidal functions in the complex domain. When trained on the large-scale VGGFace2 dataset, our model achieves 92.8% training accuracy and 91.2% validation accuracy, using only around 8.2M parameters which is substantially fewer than many state of the art approaches. These results suggest that “quantum inspired” design can enhance classical deep learning for face recognition without needing specialized quantum hardware.