Robert Stam discusses schema design in MongoDB. Some key points: - MongoDB uses collections instead of tables and documents instead of rows. - Schema design focuses on how an application will access and use data, rather than on data storage. - Relationships can be modeled through embedding documents, linking documents through IDs, or embedding IDs in other documents. The best approach depends on read/write patterns. - Examples show modeling one-to-one, one-to-many, and many-to-many relationships through embedding, linking, and linking in both directions between collections. - Categories can be modeled as documents, arrays, or hierarchical paths.