With an aim to fill the gaps in the current 3D‐printing technology to digitally fabricate medical assistive devices with significant user benefit, well‐being, and availability; we develop a design methodology that enables the optimization of lightweight multi-material lattice structures in order to enhance the design of prostheses and rehabilitation devices. This is done by firstly developing a suitable multi-variable mathematical model for topology optimization of two-scale structures and secondly demonstrating it on an outer shaft of prostheses (lower limb prostheses shaft). We develop a two-scale gradient-based optimization algorithm procedure of multiple design variables that generates functionally graded structures having excellent performance. Our design methodology employs three families of predefined micro-structures that share similar geometric features. Those two additional families thwart the convergence of our gradient-based algorithm to the global minima and we aim at presenting a computational framework that enhances multi-variable optimizations by avoiding the unfavorable local minima.