<p>Boolean functions are central to the design of secure and efficient cryptographic systems, particularly in the construction of substitution boxes (S-boxes) used within block ciphers. As security demands evolve—driven by both lightweight computing needs and the threat posed by quantum adversaries—the challenge of crafting Vectorial Boolean Functions (VBFs) that meet multiple cryptographic criteria becomes increasingly important. This paper introduces a hybrid optimization technique that combines evolutionary computation with differentiable programming to generate vBfs tailored for use in lightweight and post-quantum block cipher architectures. Unlike conventional design strategies that optimize one metric at a time, our method concurrently enhances multiple properties, including nonlinearity, differential uniformity, algebraic degree, and adherence to the strict avalanche criterion (SAC). The proposed framework generates optimized 8 × 8 S-boxes, demonstrating superior performance compared to established designs when evaluated using standard cryptographic benchmarks and spectral analyses. Moreover, our hardware simulations reveal that the resulting functions are well-suited to energy-constrained devices such as those used in IoT and embedded applications. These results underscore the practical relevance of our method for strengthening cryptographic systems that are both efficient and resistant to modern attack vectors, including those anticipated in a post-quantum security landscape.</p>