Computational design and analyses of nanoantennas obtained via surface shape optimization are presented. Starting with a kernel geometry, free deformations are applied on selected surfaces to reach optimal designs that can provide improved power enhancement capabilities at desired frequencies. An in-house implementation of genetic algorithms is efficiently combined with the multilevel fast multipole algorithm developed for accurate solutions of plasmonic problems to construct the effective optimization environment. The geometries obtained via optimization do not only represent optimal shapes within the allowed deformation limits but also reveal certain types of modifications on kernel geometries to improve their performances.