Interview Questions

What are Selenium 4 relative locators and when would you actually use them?

How do you handle multiple windows or tabs?

Capture the current handle, trigger the popup, wait for the handle count to grow, switch by diffing the handle set, then switch back to the original handle.

How do you work with iframes, and what breaks when you forget to switch back?

switchTo().frame() by WebElement (most robust), interact, then switchTo().defaultContent(); until you switch back every locator resolves inside the frame and throws NoSuchElementException.

When do you need the Actions class instead of plain click and sendKeys?

For composite input: hover menus, drag and drop, right/double click, keyboard modifiers, and click-and-hold sliders — anything the W3C actions API models as a sequence.

How do you automate file upload and verify a download?

Upload by sendKeys of an absolute path to the input[type=file] element — never open the OS dialog. For downloads, set a known download directory in browser options and poll the filesystem for the file.

Explain the Selenium 4 architecture and what changed from Selenium 3.

Selenium 4 speaks W3C WebDriver directly to the browser driver — the JSON Wire Protocol and its translation layer are gone — and Grid was rebuilt as Router, Distributor, Session Map, Queue and Node.

In what order do TestNG annotations execute?

@BeforeSuite → @BeforeTest → @BeforeClass → @BeforeMethod → @Test → @AfterMethod → @AfterClass → @AfterTest → @AfterSuite, with @BeforeGroups/@AfterGroups around grouped tests.

How do you run data-driven tests with @DataProvider, and when do you use an external file?

@DataProvider returns Object[][] (or an Iterator) that TestNG feeds into the test method once per row; move data to Excel/CSV/JSON when non-engineers own it or the set is large.

How do you configure parallel execution in TestNG?

Set parallel (methods, classes, tests or instances) plus thread-count in testng.xml, and make every shared resource thread-safe — driver in ThreadLocal, no static state, unique test data per thread.

How do you implement retry for flaky tests with IRetryAnalyzer, and when is it wrong?

Implement IRetryAnalyzer with a max-retry counter and attach it globally via an IAnnotationTransformer; use it for known infrastructure flakiness only, never to hide product bugs.

What do TestNG listeners give you, and which ones do you actually implement?

ITestListener for screenshots/logging on failure, ISuiteListener for report lifecycle, IAnnotationTransformer for retries, and IInvokedMethodListener for per-step logging.

What are the Maven lifecycle phases that matter for a test project?

validate → compile → test-compile → test → package → verify → install → deploy; running a later phase runs every earlier phase in that lifecycle.

How do dependency scopes work, and which scope should test libraries use?

compile is the default and leaks to consumers, provided is supplied at runtime by the container, runtime is not needed to compile, and test is only on the test classpath — Selenium, TestNG and REST Assured all belong in test scope.

How do you use Maven profiles to run the same suite against different environments?

Define profiles per environment that set properties (base URL, browser, suite file), then activate with -P; the framework reads the properties as system properties.

How do you run a single test, a group, or a specific suite XML from the command line?

-Dtest=ClassName#method for Surefire, -Dgroups=smoke for TestNG groups, and -Dsurefire.suiteXmlFiles or -DsuiteXmlFile to point at a suite XML.

Merge vs rebase — what do you use on an automation repository?

Rebase your own feature branch onto main to keep a linear history, merge (with a PR) into main; never rebase a branch other people have already pulled.

How do you move one fix into another branch without merging everything?

git cherry-pick the specific commit (or range) onto the target branch; for a partial file change, commit the hunk separately first so there is a clean commit to pick.

What branching strategy works for a shared automation repository?

Trunk-based: short-lived branches off main, PR review with the suite running in CI, squash merge, and release branches only when the product itself is branched.

A commit broke the suite — how do you find and undo it?

git bisect to binary-search the first bad commit, then git revert it on main; use reset only on local, unpushed history.

How does Playwright auto-waiting differ from Selenium waits?

Playwright checks actionability (attached, visible, stable, enabled, receives events) before every action and retries until the timeout, so explicit waits are rarely needed.

Which locators do you prefer in Playwright and why?

Role-based and label-based locators first (getByRole, getByLabel), then getByTestId; raw CSS/XPath last because they couple tests to markup.

How do you reuse authentication instead of logging in per test?

Log in once in a setup project, save the session with storageState, and load that state in the config so each test starts signed in.

How do you mock or intercept network calls in Playwright?

page.route intercepts matching requests so you can fulfil them with fixtures, abort them, or modify the response — ideal for error states and third-party isolation.

How do you debug a test that only fails in CI?

Turn on trace, video and screenshot on first retry, download the artifact and open the trace viewer — it replays DOM snapshots, network, console and the action timeline.

How does parallelism work in Playwright, and how do you scale it in CI?

Playwright runs files in parallel across worker processes, each with its own browser context; scale further by sharding the suite across CI machines and merging the reports.

Which HTTP status codes must an API tester know cold, and what do you assert beyond them?

200/201/202, 204, 301/302, 400, 401 vs 403, 404, 409, 422, 429, 500/502/503 — and beyond the code you assert body schema, headers, and side effects.

How do you test idempotency and retries?

Send the same request twice with the same Idempotency-Key and assert one resource is created; GET, PUT and DELETE must be naturally idempotent, POST needs a key.

How do you validate a response schema, and why is that better than field-by-field asserts?

Validate against a JSON Schema (or the OpenAPI spec) so types, required fields and enums are checked in one assertion that fails when the contract drifts.

How do you handle OAuth2 / JWT authentication in an API suite?

Fetch a token once per run through the client-credentials or password grant, cache it until shortly before expiry, and attach it via a request specification — never hardcode tokens.

How do you chain requests and manage test data for API tests?

Create prerequisites through the API itself, extract ids from the response, pass them to the next call, and clean up in teardown — no dependence on pre-seeded records.

How do you structure environments and variables in Postman?

Collection variables for defaults, environment variables per stage (dev/staging/prod), and pm.environment.set for runtime values like tokens; secrets go in the environment as secret-type variables, never in the collection.

What do you write in Postman test scripts beyond a status-code check?

Schema validation with ajv/tv4, business assertions on the parsed body, response-time budgets, header checks, and chaining values into environment variables for the next request.

How do you run Postman collections in CI?

Export the collection and environment, run them with Newman in the pipeline, publish the HTML/JUnit report, and fail the build on assertion failures.

When would you use a Postman mock server?

To unblock front-end or test development before the API exists, and to reproduce error responses that the real service will not produce on demand.

How do you avoid repeating headers and base config in REST Assured?

Build a RequestSpecification (base URI, path, content type, auth, logging) and a ResponseSpecification (status, content type, time budget) once, then reuse them in every test.

How do you serialise and deserialise payloads with POJOs?

Pass a POJO as the body and REST Assured serialises it with Jackson; call .as(Order.class) to deserialise the response into a typed object you can assert on.

How do you extract a value and chain it into the next request?

Use extract().path() or extract().jsonPath() to pull the id or token out of the response, store it in a variable, then use it in the next given().

How do you test a multipart file upload with REST Assured?

Use multiPart() with the file plus any form fields, set the content type to multipart/form-data, and assert both the response and that the stored document can be fetched back.

How do you validate deeply nested JSON without a POJO?

Use GPath through body() with Hamcrest matchers, or extract a JsonPath and assert on collections — findAll, collect and sum handle nesting cleanly.

Walk me through the layers of your automation framework.

Tests → page/service objects → core (driver factory, waits, config, data) → reporting and CI. A test contains only business steps and assertions; every technical detail lives below it.