Repository

Repository encapsulates persistence behind collection-like methods—`findById`, `save`, `delete`—so application code speaks domain language instead of SQL or ORM details. It clarifies where querying lives and enables in-memory fakes in tests. Repositories should not absorb unrelated business orchestration.

On interviews: contrast Repository with Active Record and raw DAOs. Explain aggregate-root repositories versus generic CRUD for every table.

Common pitfalls: repositories returning ORM entities everywhere; N+1 queries hidden inside convenience methods; god repositories with twenty query variants.

The trade-off is flexibility versus complexity—know when the simpler path is enough.

Checklist:

• Interface defined in application or domain layer.
• Implementation lives in infrastructure.
• Queries express domain access needs.
• Tests use in-memory or fake repositories.