Reliable materials performance has been critical for nuclear energy to make further improvements on safety and economics. The current and proposed future nuclear energy systems present an exceptionally harsh environment for structural materials due to the combination of high temperature, high stress, corrosive coolant, and intense radiation fluxes. In contrast to other energy industries, this additional radiation field brings out a number of degradation issues, and hence calls for meticulous assessment of material long-term behavior and development of advanced material resistant to degradation. In this talk, extensive computational characterization of radiation damage in materials will be demonstrated from different perspectives with multiple methods including physics-based models and data-driven techniques. There will be three coherent sections: starting with atomistic description of radiation resistance and defect interactions, then scaling up to multiscale framework to describe defect long-term evolution, and finally a holistic view using machine learning to predict radiation effects. These efforts have been aimed at a robust descriptive and predicative paradigm for materials in nuclear applications.