Primary goal of the project is to curate and enhance existing databases for adsorption-based CO₂ capture materials by correcting errors and imputing missing information using a multifidelity active learning framework. This project builds on prior work in post-combustion CO₂ capture using solid sorbents and pressure swing adsorption, where advanced machine learning and physics-informed models were developed for process design and optimization. This will be achieved by developing multifidelity active learning frameworks that integrate AI-driven screening, targeted experimentation, and process modeling to curate and improve sorbent databases. By addressing missing or inaccurate material data and jointly optimizing both material properties and process configurations through proxy models, the project will deliver a scalable solution for identifying optimal material–process combinations. This approach not only enhances CO₂ capture but also establishes a generalizable methodology for accelerating the discovery and deployment of advanced materials and processes across diverse separation challenges.