This paper introduces the HEK Shuffling algorithm, a novel variation of the Fisher-Yates Shuffling algorithm. The new algorithm offers significant performance improvements, particularly for large datasets, reducing the time complexity by 50% compared to the standard Fisher-Yates algorithm. This enhancement makes the algorithm more suitable for applications requiring efficient random shuffling of large arrays. The paper includes the theoretical foundations of the algorithm, randomness and performance analysis, the pseudocode of the algorithm, and practical use cases for general and Dropout Layer-specific applications.