Porous (or cellular) materials, like wood and bones, are abundantly found in nature for structural purposes. Thanks to the advancements of evolving technologies, the fabrication of intricate cellular structures was possible. Notably, functionally graded porous structures offer superior mechanical and physical properties compared to their uniform counterparts. Additive manufacturing, especially light-based 3D printing (3DP) technologies, has significantly augmented the production of such materials thanks to their high speed, resolution, and cost-efficiency. However, selecting the most suitable porous architecture for one's necessities remains challenging, requiring an optimised workflow supporting 3DP methods with in silico tools to predict the functional properties of the developed porous architecture. This study introduces a preliminary step toward this workflow, involving in silico design and mechanical validation of porous architectures manufactured via low-force stereolithography 3DP. The work proposes for the first time spherical-based functionally graded porous geometries, filling the existing gap in the literature. The proposed structures, categorised upon their pore size, porosity, and pore size gradients, demonstrate potential applications in the biomedical field (e.g., tissue engineering scaffolds and soft bioreactor systems) and outperform other existing porous architectures in terms of energy absorption capabilities. These findings lay the foundation for further workflow optimisation. By coupling our fast method of generation of porous structures and a trained machine learning model to predict desired physical properties, we aim to produce functionalised cellular architectures for biomedical applications.