In data modeling, what tool stack do you use to build models, and at which level of modeling do you apply normalization?

For conceptual and logical modeling I typically use ERwin or PowerDesigner, both of which support reverse engineering, forward engineering, and comprehensive ER diagramming. When working with smaller projects or quick prototypes, I prefer dbdiagram.io or Lucidchart for their simplicity and collaboration features. For the physical model, I rely on database‑specific tools such as MySQL Workbench, SQL Server Management Studio, or Oracle SQL Developer, which allow me to generate DDL scripts directly from the diagram. I apply normalization during the logical modeling phase, usually targeting Third Normal Form (3NF). This ensures that the logical schema is free of redundancy and ready for physical implementation. Once the physical model is generated, I review the design for performance considerations, adding indexes or denormalizing where necessary, but the core normalization remains 3NF.