To better understand consequences of classical assumptions regarding deformation mechanisms at the mesoscale, experimental observations of mesoscale deformation are presented. In light of actual micrographics of deformed polycrystals, the Von Mises criterion which states that 5 independent plastic deformation sources are needed at each material point to satisfy compatibility is studied, and the consequences of violating this assumption are presented through comprehensive parametric studies. From these studies, it can be concluded that not only are 5 independent plastic deformation sources not needed or observed at each point, but if less than 5 sources are allowed to be active a new physical understanding of a mechanism for kinematic hardening emerges. Furthermore, for enhanced subgrain rotation and evolution the Von Mises criterion must be violated. The second focus of this work is looking at studies, experiments, and models of mesoscale deformation in order to better understand controlling deformation length scales, so that they can be fed into a combined top-down, bottom-up, non-uniform crystal plasticity model that captures the variability provided by the mesoscale during deformation. This can in turn be used to more accurately model the heterogeneity provided by the response of each grain. The length scale intuited from insight into mesoscale deformation mechanisms through observation of experiments and analytical models is the free slip line length of each slip system, which informs non-uniform material parameters in a crystal plasticity model that control the yielding, hardening, and subsequent softening of each individual slip system. The usefulness of this non-uniform multiscale crystal plasticity model is then explored with respect to its ability to reproduce experimentally measured responses at different strain levels for different size grains. Furthermore, a "Mantle-Core" type model which combines both the non-uniform material parameter model and the limited slip model is created, in which the majority of plastic deformation is accommodated near the grain boundary under multi-slip, and uniform plastic deformation occurs in the bulk dominated by double or triple slip. These models are compared for similar levels of hardening, and the pole figures that result from their deformation are compared to experimental pole figures. While there are other models that can capture the heterogeneity introduced by mesoscale deformation at the grain scale, this combined top-down, bottom-up multiscale crystal plasticity model is by far one of the most computationally efficient as the heterogeneity of the mesoscale is does not emerge by introducing higher order terms, but rather by incorporating the heterogeneity into a simple crystal plasticity formulation. Therefore, as computational power increases, this approach will be among the first that will be able to perform accurate polycrystal level modeling while retaining the heterogeneity introduced by non-local mesoscale deformation mechanisms at the sub-grain scale.