Conventional optimization approaches for torque converters often rely on iterative trial-and-error methods, resulting in inherent limitations and uncertainties. To address these challenges and achieve performance optimization based on flow characteristics, this study proposes an inverse design method for turbomachinery. Introducing a novel index for blade loads, denoted as rvt , enables simultaneous optimization for both the cascade system and the flow field. Furthermore, a pump’s driven loads are introduced for the directional design of the blade load characteristics of high-capacity torque converters. Building upon the inverse design method, a directional optimization strategy is developed to enhance the pump capacity or torque ratio, guided by the concept of driven loads. The inverse design method allows for the targeted design of blade shapes while ensuring optimal blade shape and flow field distribution. The design results and the prototype demonstrate a significant 8.28% increase in pump capacity and a notable 4.9% reduction in the magnitude of blade loads in the internal flow field. These findings confirm the efficacy of the inverse design method in achieving directional enhancement of pump capacity while optimizing the smoothness of the internal flow field.