roadmap
- 01
- 02
- 03
- 04
- 05
- 06
- 07
- 08
Prepare for API testing interview questions on checking HTTP services, from one call to a contract. Covers status codes, Postman collections, REST Assured, Karate, service mocks and contract tests.
573 of 573 questions in your plan: 0 known, 0 in progress, 0 due, 0 skipped, 573 new. 0 cut, outside your plan.
QA and SDET loops are where API testing gets time of its own, usually as a live exercise or a take-home: here is an endpoint and its description, write the tests that would catch it breaking. Backend loops reach it inside the testing conversation. Either way the talk starts on the protocol, because an assertion's expected value is a claim about HTTP: why a client may retry a timed-out PUT but not a POST, why a missing token earns 401 and a token without the right scope earns 403.
The tool questions ask where a value comes from and what a check really proved. Which value wins when a Postman environment and its collection define the same variable, and why does a collection that passes on a laptop fail under Newman in the build? Why does plain auth().basic() in REST Assured send no credentials until the server answers with a challenge, and what does a then() that checks only the status code let through? Middle and senior candidates get the problem in reverse: a suite stayed green while production broke, and the job is to name the gap, such as a stub still returning a field the provider renamed or a contract the provider never verified.
HTTP comes before any tool, and it comes twice. Methods and status codes is the protocol's own meaning: which verbs are safe or idempotent, what each response class promises. The REST pass asks the designer's question, which verb and code an API ought to choose, with PUT, PATCH and POST the usual argument. Both lead because every tool after them only automates an expectation, and an expectation built on the wrong status code is a bug the suite will defend.
The tools follow in the order a tester tends to meet them. Postman drives a service by hand, and Newman is the turn that matters: the same collection run from a terminal, with an exit code a build can fail on. SoapUI keeps the older lineage of requests generated from a service description. REST Assured moves the checks into a JVM test suite, and Karate follows as its scenario-file alternative, with no step definitions to write and a match that compares a whole payload.
Karate's mock mode is the hinge into the last concern, the services your API calls. Service virtualization frames standing in for a dependency you cannot control, and the stub-server module makes it concrete, down to injected delays and faults. Its fidelity risks are why the plan ends where it does: a stub encodes what the consumer believes about the provider, but never tests that belief; contract testing does. A Pact consumer test records expectations, the provider replays them, and the broker turns the result into a release gate. Spring Cloud Contract is the alternative that keeps contracts in the producer's repository and generates both its tests and the consumer's stubs from them, and the module closes past request and response, on message contracts and the compatibility gates that keep a schema change from breaking readers already deployed.
Coming from manual testing or Postman alone, make the HTTP stretch solid, then the step from a saved request to a Newman run in a pipeline; an automation interview assumes you have taken it. Pick one code library by the posting's stack: REST Assured for a Java or Kotlin team, Karate where tests are written as scenarios. Backend developers usually know the protocol and are probed on stubs and contracts, so begin at the stand-ins. Senior candidates should expect the contract stretch, `can-i-deploy` and provider states in particular, to carry the conversation. By frequency, where study order opens, mixes Postman, REST Assured and Pact questions, which suits a generalist QA screen. When the posting names one tool, switch to by topics, trim importance to its must know end, and stop after that tool's module. The plan checks whether an API answers correctly, not how it holds up: nothing in it generates load or measures latency under concurrency. That is the [Grafana k6 roadmap](/roadmaps/k6). For the whole SDET loop, with a coding round and browser automation beside the API work, see the [Java SDET roadmap](/roadmaps/java-sdet).
Coming from manual testing or Postman alone, make the HTTP stretch solid, then the step from a saved request to a Newman run in a pipeline; an automation interview assumes you have taken it. Pick one code library by the posting's stack: REST Assured for a Java or Kotlin team, Karate where tests are written as scenarios. Backend developers usually know the protocol and are probed on stubs and contracts, so begin at the stand-ins. Senior candidates should expect the contract stretch, can-i-deploy and provider states in particular, to carry the conversation.
By frequency, where study order opens, mixes Postman, REST Assured and Pact questions, which suits a generalist QA screen. When the posting names one tool, switch to by topics, trim importance to its must know end, and stop after that tool's module.
The plan checks whether an API answers correctly, not how it holds up: nothing in it generates load or measures latency under concurrency. That is the Grafana k6 roadmap. For the whole SDET loop, with a coding round and browser automation beside the API work, see the Java SDET roadmap.