What is the role of the GraphQL execution engine and how are queries resolved?

Expected Interview Answer

Step-by-Step Explanation

1. Step 1

   Parse

   The incoming query string is tokenized and parsed into an abstract syntax tree (AST).

2. Step 2

   Validate

   The AST is checked against the schema — fields exist, argument types match, fragments are valid — before any resolver runs.

3. Step 3

   Build the operation

   The engine identifies the operation type (query/mutation/subscription) and its root selection set as the starting point.

4. Step 4

   Resolve top-down

   For each field it calls the resolver with (parent, args, context, info); the return value becomes the parent for nested fields.

5. Step 5

   Traverse the selection set

   Sibling fields resolve concurrently while nested fields wait for their parent, until every branch reaches scalar leaves.

6. Step 6

   Assemble the response

   Resolved values are stitched into a JSON object mirroring the query shape, with any field errors collected in the errors array.

Code Example

const resolvers = {
  Query: {
    // root field: parent is undefined here
    user: (parent, args, context) => context.db.users.findById(args.id),
  },
  User: {
    // parent is the resolved user object from Query.user
    fullName: (user) => `${user.firstName} ${user.lastName}`,
    // nested field triggers another fetch, using the parent's id
    posts: (user, args, context) => context.db.posts.byAuthor(user.id),
  },
};

// Query the engine resolves top-down:
// query { user(id: "1") { fullName posts { title } } }
// 1. Query.user runs -> returns user object
// 2. User.fullName and User.posts resolve concurrently off that object
// 3. Post.title uses the default resolver (reads post.title)

Follow-up Questions

• What are the four arguments passed to every resolver?
• How does the default resolver work when you don't write one?
• Why do top-level mutation fields execute serially instead of concurrently?
• What is the N+1 problem and how does DataLoader batch resolver calls?
• How does the info argument help with query optimization?