Abstract This study focuses on advancing Direct Air Capture (DAC) technology for atmospheric carbon dioxide reduction, employing novel polyimine materials. The study, at the intersection of molecular computation and machine learning, investigate traditional poly(ethylenimine) (PEI) sorbents and novel poly(propylenimine) (PPI) and their derivatives, addressing the pressing environmental challenge of CO2 accumulation. The research encompasses three main objectives: exploring CO2 transport in the hyperbranched PEI/MCM-41 hybrid system, in-depth investigation of CO2 transport in hyperbranched PPI membranes and corresponding MCM-41 systems, and the discovery of innovative polyimine materials for efficient CO2 capture using machine learning methodologies. Central to the dissertation is the use of Molecular Dynamics (MD) simulations for understanding CO2 capture processes within polyimine-loaded MCM-41. This involves the development of new force field parameters to accurately describe interactions between CO2, water, amine groups, and silica surfaces, alongside characterizing CO2 distribution and movement through pair correlation and mean-square displacement analyses. The study is set to provide significant insights into the enhancement of CO2 capture within these systems and contribute to the development of new materials with improved capture performance. By integrating experimental data for cross-validation and employing active learning models, the research also aims to identify new polymines exhibiting superior CO2 uptake and transport properties. This initiative is expected to substantially elevate the efficacy and practicality of DAC technology, potentially leading to reduced costs and energy demands, and aligning with global efforts towards carbon neutrality and sustainable industrial processes.