The integration of geometrical heterogeneity in structural design holds promise for meticulous tailoring of mechanical properties. Offering the possibility to design site-specific structural features, heterogeneity can serve as an additional degree of freedom, safeguarding against localized failures and averting potential catastrophic collapse. The embrace of heterogeneity in the context of structural design contributes to a more robust and adaptable framework, well-equipped to navigate mechanical challenges with heightened efficiency and durability. In this research, we try to leverage heterogeneity to design lattice structures with customized structural performance pairing numerical analysis, machine learning algorithms, and experimental tests.