Additive manufacturing enables on-demand fabrication of geometrically complex, customized components, and Laser Powder Bed Fusion (LPBF) has become a key process for high-value applications. While recent high-fidelity simulations have improved insight into LPBF melt pool formation, achieving predictive agreement with experiments is still difficult—especially when powder-scale particle dynamics interact with melt flow and when multiple constituents with distinct thermophysical properties are involved. In this work, we advance a process-scale computational mechanics capability for LPBF by developing a multi-physics CFD–DEM formulation that explicitly resolves powder spreading and particle motion, and couples these effects with ray-tracing–based energy input to capture complex deposition and irradiation scenarios [1,2]. Building on this foundation, we introduce a multi-medium, multi-physics coupled computational framework tailored to multi-material systems, enabling systematic investigation of melt pool evolution, interfacial transport, and component mixing under realistic scanning conditions [3]. To strengthen model credibility and support cross-scale interpretation, synchronous high-speed X-ray in-situ measurements are integrated with high-fidelity simulations, providing complementary observation of transient melt pool behavior and mixing phenomena in multi-material tracks. The combined experimental–computational analysis shows that multi-material melt pools develop morphologies that differ markedly from single-material cases. These differences arise from the interplay of species mixing, shallow effective thermal penetration, Marangoni-driven convection, and surface-vortex–induced recirculation, which together modify melt pool geometry and stability. Finally, by mapping the influence of scan parameters and source characteristics onto melt pool descriptors, we construct an AI-based surrogate model to accelerate prediction and guide parameter optimization for stable multi-material forming. The proposed framework provides a quantitative route to connect powder-scale physics, melt pool dynamics, and process optimization for multi-material LPBF.