This work presents SCIGEN, a machine learning framework integrating geometric constraints into material generation. The framework enables the discovery of stable quantum material candidates, and the authors synthesize two predicted magnetic materials.