Purpose of an ORM like Sequelize

WHAT THIS TESTS Whether the candidate understands the abstraction an ORM provides, its productivity benefits, and its real costs, rather than treating it as pure magic.

A GOOD ANSWER COVERS An object-relational mapper bridges the gap between relational tables and application objects. Sequelize lets you define models that map to tables, then perform CRUD through methods like findAll, create, and update instead of writing SQL strings by hand. It manages associations (hasMany, belongsTo, many-to-many through join tables), supports transactions, provides migrations and schema definitions, and parameterizes all values to defend against SQL injection. It also abstracts dialect differences across Postgres, MySQL, and others, so the same code targets multiple databases. Compared with raw pg, where you assemble query text and manage results manually, the ORM reduces boilerplate, makes relationships and eager loading declarative, and centralizes model logic.

COMMON WRONG ANSWERS Claiming the ORM is always faster than raw SQL; it adds a translation layer and can emit suboptimal queries. Saying it removes the need to understand SQL; you still need SQL to debug generated queries and tune performance. Forgetting the N+1 query problem, which ORMs make easy to trigger through lazy associations.

LIKELY FOLLOW-UPS When would you drop to raw SQL? How do you detect and fix N+1 queries with eager loading? How do migrations fit into deployment? What are the downsides of leaky abstractions?

ONE CONCRETE EXAMPLE Fetching ten authors and then their books lazily issues one query for authors plus one per author, eleven queries total, the classic N+1 problem. Using Sequelize eager loading with include collapses it into a single joined query. For a complex analytics report, the team bypassed the ORM and wrote tuned raw SQL through sequelize.query, illustrating that an ORM complements rather than replaces SQL knowledge.