This research proposed an innovative approach to sustainable architecture by developing a system for the intelligent assembly of recycled materials, which inherently vary in geometry, size, and form. The project created a comprehensive material database to catalog the properties of salvaged materials, serving as the foundation for training reinforcement learning algorithms. These algorithms optimized the assembly process, aiming for structural integrity and aesthetic quality through discrete aggregation simulations. By integrating machine learning with recycled material usage, this project not only addressed environmental sustainability by reducing construction waste but also enhanced architectural design processes, potentially leading to more innovative and cost-effective construction practices. The ultimate goal was to demonstrate that sustainable methods can be effectively aligned with architectural innovation, pushing the boundaries of technology in design.