An efficient neural operator that leverages a novel multi-grid approach as well a tensorized architecture for better performance, generalization and scalability.