In this presentation, micro- and nanoscale modelling and design of materials and structures are surveyed in view of structural and functional property relationships in Functionally Graded Materials (FGMs), thermo-electric (TE) materials, and Metal–Organic Frameworks (MOFs), through an integrated views of computational methods such as first principles modelling and machine learning (ML). In FGMs, graded composition, interfaces, and porosity control coupled thermo–mechanical–electrical–diffusive responses, where continuum or Finite-Element Models (FEMs) augmented with data-driven surrogates enable inverse design. In TE materials, multiscale mechanisms of electron and phonon transport—defects, interfaces, and nano structuring—are captured by first-principles with Boltzmann Transport Equation (BTE) or Molecular Dynamics (MD), while Crystal Graph Neural Networks (CGNN) and generative models accelerate prediction of transport coefficients and ZT optimization. For MOFs, the vast chemical/topological space is tackled by Grand Canonical Monte Carlo (GCMC) simulation and density functional theory (DFT) for adsorption/diffusion, ML models for property prediction, and synthesis-aware generative design with Bayesian optimization, enabling rapid advances in separation, storage, catalysis, and sensing. We emphasize end-to-end pipelines from first principles to features, ML/DL models, multi-scale FEMs, and device evaluation; physics-informed learning; and active integration with experimental data for reliability. Key challenges include data scarcity and bias, rigorous multiscale coupling, extrapolation and interpretability, and constraints from synthesis feasibility and durability. Also, a roadmap toward autonomous materials design via generative AI is considered under physical constraints, digital twins for design automation, and hierarchical optimization across materials–manufacturing–devices–systems.