Nanostructured materials have exceptional mechanical properties and have a wide variety of applications ranging from manufacturing, aerospace to healthcare. However, the service life of these materials is often limited due to the formation of cracks and defects that leads to their eventual failure, thus insights into atomistic mechanisms underlying these processes are valuable inputs to the design of next-generation nanostructured materials. In this talk, I will discuss large-scale molecular-dynamics simulations on leadership-scale parallel supercomputers such as ALCF-Mira, combined with data-mining and visualization techniques: (1) self-healing of cracks in alumina composite embedded with SiC/SiO2 nanoparticles, (2) ultra-light material design using nickel Kagome lattice, and (3) toughing of MoWSe2 monolayer during crack propagation by novel strain-induced structural phase transformation.