The inherent architectural complexity and stochasticity of polymeric materials combined with their myriad application domains present significant challenges for accurate data modeling and representation—hindering development of new artificial intelligence (AI) models and informatics needed to accelerate polymer research and discovery. To address these difficulties, we developed an extensible and re-composable data model for the organization of experimental data coupled with a flexible (hyper)graph representations of polymeric materials and supramolecular complexes. Implemented in a web application, these advances enabled the curation of more than 2600 experimental records across disparate experiment types, including small-molecule organic synthesis, polymer synthesis, supramolecular complex synthesis, and the assay of antimicrobial materials. The unified dataset enabled creation of chem- and polymer informatics and generative models for new ring-opening polymerization catalysts. These results demonstrate how an extensible data model and flexible polymer representation enable a customizable approach for unifying experimental materials data across multiple domains and provides a critical link to merge AI and automation in accelerated discovery workflows.