One of the major goal of the researchers is to reduce energy loss including nanoscale to the structural level. For instance, around 65% of fuel energy is lost during the propulsion of the automobiles, where 11% of the loss happens at tires due to rolling friction. Out of that tire loss, 90 to 95% loss happens due to hysteresis of tire materials. This dissertation focuses on multiscale modeling techniques in order to facilitate the discovery new rubber materials. Enhanced coarse-grained models of elastomers (thermoplastic polyurethane elastomer and natural rubber) are constructed from full-atomic models with reasonable repeat units/beads associated with pressure-correction for non-bonded interactions of the beads using inverse Boltzmann method (IBM). Equivalent continuum modeling is performed with volumetric/isochoric loading to predict macroscopic mechanical properties using molecular mechanics (MM) and molecular dynamics (MD). Glass-transition and rate-dependent mechanical properties along with hysteresis loss under uniaxial deformation is predicted with varying composition of the material. A statistical non-Gaussian treatment of a rubber chain is performed and linked with molecular dynamics in order predict hyperelastic material constants without fitting with any experimental data.