In the latest FAO projection, the global population is expected to reach 10 billion by 2050. Thus, enhancing agricultural productivity while minimizing ecological impact has become an urgent priority. Meeting this challenge demands innovative strategies that promote sustainable land management, soil conservation, and efficient water usage to ensure food security. Soilless cultivation methods, such as hydroponics, aeroponics, and aquaponics practices, can address these challenges through sustainable alternatives to conventional agriculture. These systems rely on porous organic or inorganic substrates (e.g., peat moss, rockwool, expanded clay), and/or on artificial supporting structures. Depending on the selected practice, material and/or structure, different structural, hydraulic, water retention, and porosity properties can be tailored to specific needs. Inspired by the efficiency of natural porous microstructures (e.g., sponges, bones, wood), we developed an engineered framework to design metamaterials in order to produce optimized 3D-printed porous scaffolds for soilless cultivations. The proposed approach employs advanced mathematical models and numerical methods, integrating multi-objective and multi-physics topology optimization (TO) with homogenization and dehomogenization techniques. Specifically, we combine a density-based TO technique to shape optimized unit cells with asymptotic inverse homogenization theory to link macroscopic properties with periodic microstructures topology. This process is further enhanced by a novel anisotropic mesh adaptation technique, enabling precise handling of periodic boundary conditions. Finally, once the optimized unitary cells are obtained, a dehomogenization phase is invoked to reconstruct the macroscopic porous scaffold. The presentation customizes the proposed workflow to the design of porous substrates, by proving the effectiveness of the procedure in replicating the behavior of natural growing media. Through numerical test cases, we highlight the potential of mathematical models and numerical methods to drive innovation in agricultural engineering.