Shape memory polymers (SMPs) are a versatile class of smart materials widely employed across several application domains due to their ability to recover a programmed shape from a temporary configuration in response to external stimuli. Beyond the conventional one-way shape memory effect, two-way SMPs exhibit reversible, load-free shape changes, making them particularly attractive for actuation. Nevertheless, achieving controlled, tunable, and application-specific actuation remains a major open challenge. This work presents recent advances in the development of novel actuation mechanisms for thermo-responsive two-way SMPs, with a focus on multi-phase chemically-crosslinked semi-crystalline polymer networks. We show how tunable actuation can be rationally achieved by leveraging three ingredients within a unified theoretical and computational framework: (i) crystallization kinetics, (ii) photo-crosslinking parameters, and (iii) architected material designs. These mechanisms are further integrated with additive manufacturing to realize 4D printed systems with complex geometries. First, a finite-strain continuum modeling framework is introduced to describe the thermo-mechanical behavior and different types of two-way shape memory effects in semi-crystalline networks. Both thermally driven and isothermal crystallization kinetics are incorporated, enabling systematic tailoring of actuation behavior. Model predictions are validated against experimental data on homo- and copolymer networks [JMPS, 195, 105955, 2025]. The framework is then extended to include a physics-based description of the photo-crosslinking process, providing new insight into polymer network formation and its impact on macroscopic actuation, and validated through a dedicated experimental campaign [MARC, e00631, 2025]. Finally, an extrusion-based 4D printing strategy [Mat. & Des., 238, 112725, 2024] is employed to design architected SMP systems with programmable mechanical and actuation properties. The interplay between material behavior, geometric nonlinearity, and structural instabilities is discussed through representative case studies. The results highlight the potential of integrated modeling and manufacturing approaches for next-generation programmable actuation systems based on SMPs. Acknowledgments. G. Scalet acknowledges that this work was funded by the European Union ERC CoDe4Bio Grant ID 101039467.