Crystal plasticity models are increasingly used in engineering applications to obtain microstructure-sensitive mechanical response of polycrystalline materials. Three key elements are: a proper consideration of the single crystal plastic deformation mechanisms, a representative description of the microstructure, and an appropriate scheme to connect the microstates with the macroscopic response. The latter can be based on homogenization (e.g. self-consistent methods), relying on a statistical description of the microstructure, or full-field solutions, requiring a spatial description of the microstructure (e.g. FFT-based methods). Full-field models are computationally intensive, preventing their direct embedding in multiscale calculations, but can be used to generate reference solutions for assessment of homogenization approaches. In this talk we will review our recent efforts to develop material models--with emphasis on algorithmic aspects--based on polycrystal plasticity to capture anisotropic mechanical behavior, along with their embedding in FE to solve problems involving complex geometries and boundary conditions with microstructure-sensitive material response.