<p>Predictive fabrication of superhydrophobic surfaces via laser processing remains a key challenge in materials science. To address this, the present study proposes and validates a closed-loop research framework that integrates theoretical modeling, intelligent optimization, and simulation-based validation, aiming to achieve precise control from microstructural design to macroscopic performance. A dynamic wetting model based on energy conservation is developed, incorporating the hierarchical roughness unique to laser processing. This model provides theoretical insight into how surface geometries influence droplet spreading dynamics. Moreover, to bridge the gap between theoretical design and practical fabrication, a data-driven optimization strategy is employed. A deep neural network (DNN) is used as a high-fidelity surrogate model, combined with the NSGA-II genetic algorithm to efficiently search the high-dimensional parameter space. The framework successfully predicted and guided the fabrication of a NiTi-based superhydrophobic surface with a static contact angle of 162.6° and a sliding angle of 4.3°, validating its effectiveness. Multiphysics numerical simulations are conducted to investigate the dynamic response of droplets on heterogeneous wetting surfaces. Contrary to intuitive expectations, surfaces with a hydrophilic core and hydrophobic periphery are found to more effectively suppress post-impact droplet oscillation compared to the reverse configuration, offering a novel strategy for controlling droplet dynamics via surface patterning. Overall, this study establishes and verifies a comprehensive, performance-driven design-to-manufacturing methodology, providing a systematic solution for the development of functional surfaces with tailored dynamic wettability.</p> Graphical abstract <p></p>