SoftwareTestPilot
28 SDET Q&A

SDET Interview Questions & Answers (2026)

28 real SDET interview questions with senior-QA answers — coding, OOP, framework design, Selenium/Playwright, REST Assured, CI/CD sharding, and system-design scenarios.

  • 5 min read
  • Difficulty: Mixed (Easy → Hard)
  • Freshers → 8+ yrs
  • Updated July 2026
  • Avinash Kamble
0 / 28 reviewed
0%

1. SDET Role & Fundamentals

Easy Very Common 1 minQ1 / 28

Q1.What is an SDET and how is it different from a manual QA?

Asked byNetflixMetaAtlassianUber
Why interviewers ask this

Interviewers open with "SDET and how is it different from a manual QA" to confirm you can define the concept in one crisp line before going deeper. In SDET rounds this filters out candidates who only remember syntax and can't articulate the underlying idea to a non-expert teammate.

Detailed explanation

An SDET (Software Development Engineer in Test) writes production-grade automation, tooling, and CI infrastructure — not just test cases. Compared with a manual QA, an SDET codes daily in Java/Python/TypeScript, owns the framework, builds test data services, and is measured on flaky-rate, coverage, and pipeline speed.

Tips to remember
  • Open with a one-sentence definition of SDET and how is it different from a manual QA, then a concrete SDET example — never start with history or theory.
  • Keep the answer to 60–90 seconds; anything longer signals you can't summarise SDET and how is it different from a manual QA cleanly.
  • Quote a real flake-rate number before and after your fix; measured outcomes score far higher than "we added retries".
RelatedQ3
Easy Very Common 1 minQ2 / 28

Q2.What does a day in the life of an SDET look like?

Asked byMicrosoftNetflixMetaAtlassian
Why interviewers ask this

This SDET question checks whether you can go beyond textbook knowledge on What does a day in the life of an SDET look like and reason about it the way a working QA engineer does — with a definition, an example, and the edge case that usually comes up next.

Detailed explanation
  • Review PRs and add regression coverage
  • Debug flaky tests in the CI dashboard
  • Pair with devs on testability (hooks, IDs, test endpoints)
  • Extend the framework (new reporter, retry logic, sharding)
  • Own release sign-off gates and quality metrics
Tips to remember
  • Anchor the answer in a real SDET project — panels reward specificity on What does a day in the life of an SDET look like over textbook wording.
  • Group the What does a day in the life of an SDET look like points into 2–3 buckets so you can recall them under pressure without missing one.
  • Quote a real flake-rate number before and after your fix; measured outcomes score far higher than "we added retries".
Easy Very Common 1 minQ3 / 28

Q3.What skills are non-negotiable for an SDET in 2026?

Asked byAtlassianUberGoogleSalesforce
Why interviewers ask this

This SDET question checks whether you can go beyond textbook knowledge on What skills are non-negotiable for an SDET in 2026 and reason about it the way a working QA engineer does — with a definition, an example, and the edge case that usually comes up next.

Detailed explanation

Strong OOP + one language (Java, Python, or TypeScript), Selenium/Playwright, REST Assured or requests, SQL, Git, Docker, one CI (GitHub Actions/Jenkins), AWS basics, and design patterns (POM, Factory, Builder). AI-assisted test authoring (Copilot, Cursor) is now a table-stakes skill.

Tips to remember
  • Anchor the answer in a real SDET project — panels reward specificity on What skills are non-negotiable for an SDET in 2026 over textbook wording.
  • Keep the answer to 60–90 seconds; anything longer signals you can't summarise What skills are non-negotiable for an SDET in 2026 cleanly.
  • Add one sentence on what does NOT belong in a page object (assertions, test data) — that boundary is what separates mid from senior answers.
Easy Very Common 1 minQ4 / 28

Q4.What is the testing pyramid and why does an SDET care?

Asked byMetaAtlassianUberGoogle
Why interviewers ask this

Interviewers open with "testing pyramid and why does an SDET care" to confirm you can define the concept in one crisp line before going deeper. In SDET rounds this filters out candidates who only remember syntax and can't articulate the underlying idea to a non-expert teammate.

Detailed explanation

Unit (70%) → Integration/API (20%) → UI/E2E (10%). SDETs push coverage down the pyramid because unit and API tests are 100× faster and less flaky than UI. In interviews, cite this when asked "how would you cover X?".

Tips to remember
  • Open with a one-sentence definition of testing pyramid and why does an SDET care, then a concrete SDET example — never start with history or theory.
  • Keep the answer to 60–90 seconds; anything longer signals you can't summarise testing pyramid and why does an SDET care cleanly.
  • Quote a real flake-rate number before and after your fix; measured outcomes score far higher than "we added retries".
Easy Very Common 1 minQ5 / 28

Q5.How do you measure automation ROI?

Asked byGoogleSalesforceAmazonMicrosoft
Why interviewers ask this

Hands-on "how would you measure automation ROI" questions reveal whether you've actually shipped SDET code or only read about it. Interviewers listen for concrete steps, the tools you'd reach for first, and the failure mode you'd guard against.

Detailed explanation

(Time saved per run × runs per week) − (authoring + maintenance cost). Also track: flaky rate < 1%, mean time to detect regression, and % of PRs blocked by real bugs vs false failures.

Tips to remember
  • Walk through measure automation ROI as numbered steps and call out the tool, command, or API used at each step.
  • Keep the answer to 60–90 seconds; anything longer signals you can't summarise measure automation ROI cleanly.
  • Quote a real flake-rate number before and after your fix; measured outcomes score far higher than "we added retries".
Confidence check

If you can confidently answer the SDET Role & Fundamentals questions above, you're well prepared for this section of your interview. Move on, or rehearse the trickier ones aloud with our AI mock interviewer.

2. Coding, DSA & OOP

Medium Very Common 1 minQ6 / 28

Q6.Reverse a string without built-ins (Java).

Asked byUberGoogleSalesforceAmazon
Why interviewers ask this

This SDET question checks whether you can go beyond textbook knowledge on Reverse a string without built-ins (Java) and reason about it the way a working QA engineer does — with a definition, an example, and the edge case that usually comes up next.

Detailed explanation
public String reverse(String s) {
  char[] a = s.toCharArray();
  int i = 0, j = a.length - 1;
  while (i < j) { char t = a[i]; a[i++] = a[j]; a[j--] = t; }
  return new String(a);
}
Tips to remember
  • Anchor the answer in a real SDET project — panels reward specificity on Reverse a string without built-ins (Java) over textbook wording.
  • Be ready to whiteboard the Reverse a string without built-ins (Java) snippet live — panels often ask you to type it, not describe it.
  • Frame the answer against the test pyramid and CI/CD feedback loop.
Medium Very Common 1 minQ7 / 28

Q7.Find the first non-repeating character (Python).

Asked byAmazonMicrosoftNetflixMeta
Why interviewers ask this

This SDET question checks whether you can go beyond textbook knowledge on Find the first non-repeating character (Python) and reason about it the way a working QA engineer does — with a definition, an example, and the edge case that usually comes up next.

Detailed explanation
from collections import Counter
def first_unique(s):
    c = Counter(s)
    return next((ch for ch in s if c[ch] == 1), None)
Tips to remember
  • Anchor the answer in a real SDET project — panels reward specificity on Find the first non-repeating character (Python) over textbook wording.
  • Be ready to whiteboard the Find the first non-repeating character (Python) snippet live — panels often ask you to type it, not describe it.
  • Frame the answer against the test pyramid and CI/CD feedback loop.
Medium Very Common 1 minQ8 / 28

Q8.Explain OOP pillars with a test-framework example.

Asked bySalesforceAmazonMicrosoftNetflix
Why interviewers ask this

Open-ended "explain OOP pillars with a test-framework example" prompts test how you structure a technical answer under pressure. Panels look for a clear opening definition, one worked example, and a closing sentence on the pitfall they were about to ask about next.

Detailed explanation
  • Encapsulation — private WebDriver inside BasePage
  • Inheritance — LoginPage extends BasePage
  • Polymorphism — overloaded click(By) vs click(WebElement)
  • Abstraction — page methods hide locators from tests
Tips to remember
  • Use a 3-part frame for OOP pillars with a test-framework example: what it is → how it works → one gotcha you've hit in a real SDET project.
  • Be ready to whiteboard the OOP pillars with a test-framework example snippet live — panels often ask you to type it, not describe it.
  • Frame the answer against the test pyramid and CI/CD feedback loop.
Medium Very Common 1 minQ9 / 28

Q9.Difference between abstract class and interface (Java 17+).

Asked byNetflixMetaAtlassianUber
Why interviewers ask this

Comparison questions like this test whether you understand abstract class vs interface (Java 17+) at a design level — not just that both exist, but when to pick one over the other. Panels use it to see if you can defend a trade-off with a real project example.

Detailed explanation

Interface = pure contract, can have default/static methods, multiple-inherited. Abstract class = partial implementation with state, single-inherited. Use interface for capability (Loadable, Retryable); abstract class for shared base (BasePage).

Tips to remember
  • Structure the answer as a small table in your head: dimension, option A, option B — and close with "I'd pick X when Y".
  • Be ready to whiteboard the abstract class vs interface (Java 17+) snippet live — panels often ask you to type it, not describe it.
  • Tie contract checks to the consumer/provider workflow and where the contract is stored — that's the senior detail panels wait for.
Easy Very Common 1 minQ10 / 28

Q10.What are Big-O tradeoffs for HashMap vs TreeMap?

Asked byMicrosoftNetflixMetaAtlassian
Why interviewers ask this

Interviewers open with "Big-O tradeoffs for HashMap vs TreeMap" to confirm you can define the concept in one crisp line before going deeper. In SDET rounds this filters out candidates who only remember syntax and can't articulate the underlying idea to a non-expert teammate.

Detailed explanation

HashMap: average O(1) get/put, no ordering. TreeMap: O(log n), sorted by key. For test-data lookups use HashMap; for range queries (e.g. "all builds after date X") use TreeMap.

Tips to remember
  • Open with a one-sentence definition of Big-O tradeoffs for HashMap vs TreeMap, then a concrete SDET example — never start with history or theory.
  • Keep the answer to 60–90 seconds; anything longer signals you can't summarise Big-O tradeoffs for HashMap vs TreeMap cleanly.
  • Frame the answer against the test pyramid and CI/CD feedback loop.
Medium Very Common 1 minQ11 / 28

Q11.Implement a rate-limited retry (TypeScript).

Asked byAtlassianUberGoogleSalesforce
Why interviewers ask this

This SDET question checks whether you can go beyond textbook knowledge on Implement a rate-limited retry (TypeScript) and reason about it the way a working QA engineer does — with a definition, an example, and the edge case that usually comes up next.

Detailed explanation
async function retry<T>(fn:()=>Promise<T>, tries=3, delay=500):Promise<T>{
  let last:unknown;
  for(let i=0;i<tries;i++){
    try{ return await fn(); }
    catch(e){ last=e; await new Promise(r=>setTimeout(r, delay*2**i)); }
  }
  throw last;
}
Tips to remember
  • Anchor the answer in a real SDET project — panels reward specificity on Implement a rate-limited retry (TypeScript) over textbook wording.
  • Be ready to whiteboard the Implement a rate-limited retry (TypeScript) snippet live — panels often ask you to type it, not describe it.
  • Frame the answer against the test pyramid and CI/CD feedback loop.
Confidence check

If you can confidently answer the Coding, DSA & OOP questions above, you're well prepared for this section of your interview. Move on, or rehearse the trickier ones aloud with our AI mock interviewer.

3. UI & API Automation

Easy Common 1 minQ12 / 28

Q12.Selenium vs Playwright — which would you pick and why?

Asked byMetaAtlassianUberGoogle
Why interviewers ask this

Comparison questions like this test whether you understand Selenium vs Playwright — which would you pick and why at a design level — not just that both exist, but when to pick one over the other. Panels use it to see if you can defend a trade-off with a real project example.

Detailed explanation

Playwright for greenfield in 2026 — auto-wait, network mocking, parallel isolation, trace viewer. Selenium when the org has a mature Java+Grid stack or needs BiDi across every browser vendor. Both are valid; interviewers want the reasoning, not the fanboy answer.

Tips to remember
  • Structure the answer as a small table in your head: dimension, option A, option B — and close with "I'd pick X when Y".
  • Keep the answer to 60–90 seconds; anything longer signals you can't summarise Selenium vs Playwright — which would you pick and why cleanly.
  • State that you never mix wait strategies and never use a hard sleep, then give the timeout value you actually run in CI.
Medium Common 1 minQ13 / 28

Q13.How do you make UI tests non-flaky?

Asked byGoogleSalesforceAmazonMicrosoft
Why interviewers ask this

Hands-on "how would you make UI tests non-flaky" questions reveal whether you've actually shipped SDET code or only read about it. Interviewers listen for concrete steps, the tools you'd reach for first, and the failure mode you'd guard against.

Detailed explanation
  • Use role/text selectors, not XPath by index
  • Wait on state (visibility, network idle), never sleep
  • Isolate test data per test (API seeding + teardown)
  • Retry only the transport layer, never the assertion
  • Run in Docker with fixed viewport & timezone
Tips to remember
  • Walk through make UI tests non-flaky as numbered steps and call out the tool, command, or API used at each step.
  • Be ready to whiteboard the make UI tests non-flaky snippet live — panels often ask you to type it, not describe it.
  • Say explicitly that you prefer CSS/relative locators and only fall back to XPath for text or axes — panels grade locator hygiene here.
Medium Common 1 minQ14 / 28

Q14.How do you handle authentication once for 200 tests?

Asked byUberGoogleSalesforceAmazon
Why interviewers ask this

Hands-on "how would you handle authentication once for 200 tests" questions reveal whether you've actually shipped SDET code or only read about it. Interviewers listen for concrete steps, the tools you'd reach for first, and the failure mode you'd guard against.

Detailed explanation

Log in via API, save the storage state / cookies to a file, and inject into every browser context. In Playwright: storageState: 'auth.json'. In Selenium: add cookies after driver init. Saves ~200 × login latency.

Tips to remember
  • Walk through handle authentication once for 200 tests as numbered steps and call out the tool, command, or API used at each step.
  • Be ready to whiteboard the handle authentication once for 200 tests snippet live — panels often ask you to type it, not describe it.
  • Cover token expiry and refresh in your answer; most candidates only describe the happy-path login.
Easy Common 1 minQ15 / 28

Q15.What is API contract testing and which tool would you use?

Asked byAmazonMicrosoftNetflixMeta
Why interviewers ask this

Interviewers open with "API contract testing and which tool would you use" to confirm you can define the concept in one crisp line before going deeper. In SDET rounds this filters out candidates who only remember syntax and can't articulate the underlying idea to a non-expert teammate.

Detailed explanation

Contract testing verifies request/response shape between services (Pact, Spring Cloud Contract). It catches breaking changes before integration. SDETs own the consumer contracts and publish them to the broker in CI.

Tips to remember
  • Open with a one-sentence definition of API contract testing and which tool would you use, then a concrete SDET example — never start with history or theory.
  • Keep the answer to 60–90 seconds; anything longer signals you can't summarise API contract testing and which tool would you use cleanly.
  • Tie contract checks to the consumer/provider workflow and where the contract is stored — that's the senior detail panels wait for.
Medium Common 1 minQ16 / 28

Q16.Write a REST Assured test for a POST + assertion.

Asked bySalesforceAmazonMicrosoftNetflix
Why interviewers ask this

This SDET question checks whether you can go beyond textbook knowledge on Write a REST Assured test for a POST + assertion and reason about it the way a working QA engineer does — with a definition, an example, and the edge case that usually comes up next.

Detailed explanation
given().contentType(JSON).body(payload)
.when().post("/orders")
.then().statusCode(201)
       .body("id", notNullValue())
       .body("status", equalTo("PENDING"));
Tips to remember
  • Anchor the answer in a real SDET project — panels reward specificity on Write a REST Assured test for a POST + assertion over textbook wording.
  • Be ready to whiteboard the Write a REST Assured test for a POST + assertion snippet live — panels often ask you to type it, not describe it.
  • Assert status, schema and business payload — naming all three signals you test contracts, not just happy paths.
Medium Common 1 minQ17 / 28

Q17.How do you test a WebSocket / real-time feature?

Asked byNetflixMetaAtlassianUber
Why interviewers ask this

Hands-on "how would you test a WebSocket / real-time feature" questions reveal whether you've actually shipped SDET code or only read about it. Interviewers listen for concrete steps, the tools you'd reach for first, and the failure mode you'd guard against.

Detailed explanation

Use a WS client library (ws in Node, okhttp WebSocket in Java). Connect, send message, assert on received frames within a timeout. For UI, mock the WS in Playwright with page.route or a local ws server.

Tips to remember
  • Walk through test a WebSocket / real-time feature as numbered steps and call out the tool, command, or API used at each step.
  • Be ready to whiteboard the test a WebSocket / real-time feature snippet live — panels often ask you to type it, not describe it.
  • Say when you'd mock versus hit the real dependency; unconditional mocking is a red flag for integration coverage.
Medium Common 1 minQ18 / 28

Q18.How do you test file upload / download E2E?

Asked byMicrosoftNetflixMetaAtlassian
Why interviewers ask this

Hands-on "how would you test file upload / download E2E" questions reveal whether you've actually shipped SDET code or only read about it. Interviewers listen for concrete steps, the tools you'd reach for first, and the failure mode you'd guard against.

Detailed explanation

Upload: setInputFiles in Playwright or sendKeys(path) in Selenium. Download: set browser download dir, wait for file, hash-compare against expected. Never rely on native OS dialogs.

Tips to remember
  • Walk through test file upload / download E2E as numbered steps and call out the tool, command, or API used at each step.
  • Be ready to whiteboard the test file upload / download E2E snippet live — panels often ask you to type it, not describe it.
  • Frame the answer against the test pyramid and CI/CD feedback loop.
Confidence check

If you can confidently answer the UI & API Automation questions above, you're well prepared for this section of your interview. Move on, or rehearse the trickier ones aloud with our AI mock interviewer.

4. Framework Design & CI/CD

Easy Common 1 minQ19 / 28

Q19.Explain your framework layers.

Asked byAtlassianUberGoogleSalesforce
Why interviewers ask this

Open-ended "explain your framework layers" prompts test how you structure a technical answer under pressure. Panels look for a clear opening definition, one worked example, and a closing sentence on the pitfall they were about to ask about next.

Detailed explanation
  1. Tests (business language)
  2. Page Objects / API clients
  3. Core (driver factory, config, waits, retry)
  4. Data (builders, fixtures, factories)
  5. Reporting (Allure/HTML + Slack)
Tips to remember
  • Use a 3-part frame for your framework layers: what it is → how it works → one gotcha you've hit in a real SDET project.
  • Group the your framework layers points into 2–3 buckets so you can recall them under pressure without missing one.
  • Add one sentence on what does NOT belong in a page object (assertions, test data) — that boundary is what separates mid from senior answers.
Medium Common 1 minQ20 / 28

Q20.How do you design test data for parallel runs?

Asked byMetaAtlassianUberGoogle
Why interviewers ask this

Hands-on "how would you design test data for parallel runs" questions reveal whether you've actually shipped SDET code or only read about it. Interviewers listen for concrete steps, the tools you'd reach for first, and the failure mode you'd guard against.

Detailed explanation

Each test creates its own data via API with a unique run-id in the payload (e.g. user_{uuid{'}'}@qa.com). No shared golden accounts. Cleanup in @AfterEach or nightly job.

Tips to remember
  • Walk through design test data for parallel runs as numbered steps and call out the tool, command, or API used at each step.
  • Be ready to whiteboard the design test data for parallel runs snippet live — panels often ask you to type it, not describe it.
  • Call out test isolation and shared-state risk (data, sessions, ports) before you talk about worker counts.
Medium Occasional 1 minQ21 / 28

Q21.How do you shard 1,000 tests across a CI matrix?

Asked byGoogleSalesforceAmazonMicrosoft
Why interviewers ask this

Hands-on "how would you shard 1,000 tests across a CI matrix" questions reveal whether you've actually shipped SDET code or only read about it. Interviewers listen for concrete steps, the tools you'd reach for first, and the failure mode you'd guard against.

Detailed explanation

Split by file count and historical duration (weighted sharding). Playwright: --shard=1/4. GitHub Actions: strategy.matrix.shard: [1,2,3,4] with fail-fast: false. Merge reports at the end via a "report" job that downloads artifacts.

Tips to remember
  • Walk through shard 1,000 tests across a CI matrix as numbered steps and call out the tool, command, or API used at each step.
  • Be ready to whiteboard the shard 1,000 tests across a CI matrix snippet live — panels often ask you to type it, not describe it.
  • Describe the pipeline stage order and what makes the build fail — a vague "we run tests in CI" answer stalls here.
Easy Occasional 1 minQ22 / 28

Q22.How do you gate a PR on quality?

Asked byUberGoogleSalesforceAmazon
Why interviewers ask this

Hands-on "how would you gate a PR on quality" questions reveal whether you've actually shipped SDET code or only read about it. Interviewers listen for concrete steps, the tools you'd reach for first, and the failure mode you'd guard against.

Detailed explanation

PR must pass: lint → unit → API → smoke UI (≤5 min). Full E2E runs on merge to main. Block merge on flaky-quarantine rate > 2% or coverage drop > 1%.

Tips to remember
  • Walk through gate a PR on quality as numbered steps and call out the tool, command, or API used at each step.
  • Keep the answer to 60–90 seconds; anything longer signals you can't summarise gate a PR on quality cleanly.
  • Quote a real flake-rate number before and after your fix; measured outcomes score far higher than "we added retries".
Medium Occasional 1 minQ23 / 28

Q23.How do you integrate visual regression?

Asked byAmazonMicrosoftNetflixMeta
Why interviewers ask this

Hands-on "how would you integrate visual regression" questions reveal whether you've actually shipped SDET code or only read about it. Interviewers listen for concrete steps, the tools you'd reach for first, and the failure mode you'd guard against.

Detailed explanation

Percy/Applitools/Playwright toHaveScreenshot(). Baseline per browser+viewport, mask dynamic regions (dates, avatars), review diffs in PR. Keep visual tests to critical pages only — they are expensive.

Tips to remember
  • Walk through integrate visual regression as numbered steps and call out the tool, command, or API used at each step.
  • Be ready to whiteboard the integrate visual regression snippet live — panels often ask you to type it, not describe it.
  • Say who reads the report and what decision it drives — evidence-for-humans framing beats a tool name list.
Easy Occasional 1 minQ24 / 28

Q24.How do you monitor test health in production pipelines?

Asked bySalesforceAmazonMicrosoftNetflix
Why interviewers ask this

Hands-on "how would you monitor test health in production pipelines" questions reveal whether you've actually shipped SDET code or only read about it. Interviewers listen for concrete steps, the tools you'd reach for first, and the failure mode you'd guard against.

Detailed explanation
  • Track pass rate, flaky rate, duration per suite
  • Auto-quarantine tests failing 3× in a row
  • Ship weekly quality report to eng leadership
  • Alert on suite duration regressing > 20%
Tips to remember
  • Walk through monitor test health in production pipelines as numbered steps and call out the tool, command, or API used at each step.
  • Group the monitor test health in production pipelines points into 2–3 buckets so you can recall them under pressure without missing one.
  • Quote a real flake-rate number before and after your fix; measured outcomes score far higher than "we added retries".
Confidence check

If you can confidently answer the Framework Design & CI/CD questions above, you're well prepared for this section of your interview. Move on, or rehearse the trickier ones aloud with our AI mock interviewer.

5. System Design & Scenarios

Easy Occasional 1 minQ25 / 28

Q25.Design a test-automation framework from scratch.

Asked byNetflixMetaAtlassianUber
Why interviewers ask this

This SDET question checks whether you can go beyond textbook knowledge on Design a test-automation framework from scratch and reason about it the way a working QA engineer does — with a definition, an example, and the edge case that usually comes up next.

Detailed explanation

Pick language matching devs, Playwright/Selenium for UI, REST Assured/axios for API, POM + fixtures pattern, Allure report, GitHub Actions with sharding, Docker for reproducibility, secrets in vault, storage-state auth, quarantine label, Slack notifier. Draw the diagram — interviewers score the drawing more than the words.

Tips to remember
  • Anchor the answer in a real SDET project — panels reward specificity on Design a test-automation framework from scratch over textbook wording.
  • Keep the answer to 60–90 seconds; anything longer signals you can't summarise Design a test-automation framework from scratch cleanly.
  • Add one sentence on what does NOT belong in a page object (assertions, test data) — that boundary is what separates mid from senior answers.
Easy Occasional 1 minQ26 / 28

Q26.How would you test an e-commerce checkout end-to-end?

Asked byMicrosoftNetflixMetaAtlassian
Why interviewers ask this

Hands-on "how would you test an e-commerce checkout end-to-end" questions reveal whether you've actually shipped SDET code or only read about it. Interviewers listen for concrete steps, the tools you'd reach for first, and the failure mode you'd guard against.

Detailed explanation
  1. API: seed user + product
  2. UI: login via storage-state, add to cart, checkout with stubbed payment
  3. DB: assert order row + inventory decrement
  4. Webhook: assert order-created event to downstream
Tips to remember
  • Walk through test an e-commerce checkout end-to-end as numbered steps and call out the tool, command, or API used at each step.
  • Group the test an e-commerce checkout end-to-end points into 2–3 buckets so you can recall them under pressure without missing one.
  • Say when you'd mock versus hit the real dependency; unconditional mocking is a red flag for integration coverage.
Easy Occasional 1 minQ27 / 28

Q27.How do you test an AI/LLM feature?

Asked byAtlassianUberGoogleSalesforce
Why interviewers ask this

Hands-on "how would you test an AI/LLM feature" questions reveal whether you've actually shipped SDET code or only read about it. Interviewers listen for concrete steps, the tools you'd reach for first, and the failure mode you'd guard against.

Detailed explanation

Split deterministic vs non-deterministic. Deterministic: prompt→schema validation, latency SLO, guardrail regexes. Non-deterministic: eval sets with LLM-as-judge, golden Q&A with cosine similarity threshold, red-team prompt-injection suite.

Tips to remember
  • Walk through test an AI/LLM feature as numbered steps and call out the tool, command, or API used at each step.
  • Keep the answer to 60–90 seconds; anything longer signals you can't summarise test an AI/LLM feature cleanly.
  • Tie contract checks to the consumer/provider workflow and where the contract is stored — that's the senior detail panels wait for.
Easy Occasional 1 minQ28 / 28

Q28.Tell me about a bug you found that saved production.

Asked byMetaAtlassianUberGoogle
Why interviewers ask this

This SDET question checks whether you can go beyond textbook knowledge on Tell me about a bug you found that saved production and reason about it the way a working QA engineer does — with a definition, an example, and the edge case that usually comes up next.

Detailed explanation

Use STAR: Situation (release week), Task (own smoke suite), Action (spotted a race between refresh-token and API-idle), Result (blocked release, patched, added contract test). Quantify: users saved, dollars, minutes downtime avoided.

Tips to remember
  • Anchor the answer in a real SDET project — panels reward specificity on Tell me about a bug you found that saved production over textbook wording.
  • Keep the answer to 60–90 seconds; anything longer signals you can't summarise Tell me about a bug you found that saved production cleanly.
  • Tie contract checks to the consumer/provider workflow and where the contract is stored — that's the senior detail panels wait for.
RelatedQ26
Confidence check

If you can confidently answer the System Design & Scenarios questions above, you're well prepared for this section of your interview. Move on, or rehearse the trickier ones aloud with our AI mock interviewer.

Quick revision

  1. Q1: What is an SDET and how is it different from a manual QA — An SDET (Software Development Engineer in Test) writes production-grade automation, tooling, and CI infrastructure — not just test cases.
  2. Q2: What does a day in the life of an SDET look like — Review PRs and add regression coverage Debug flaky tests in the CI dashboard Pair with devs on testability (hooks, IDs, test endpoints) Extend the framework (new reporter, retry lo
  3. Q3: What skills are non-negotiable for an SDET in 2026 — Strong OOP + one language (Java, Python, or TypeScript), Selenium/Playwright, REST Assured or requests, SQL, Git, Docker, one CI (GitHub Actions/Jenkins), AWS basics, and design pa
  4. Q4: What is the testing pyramid and why does an SDET care — Unit (70%) → Integration/API (20%) → UI/E2E (10%).
  5. Q5: How do you measure automation ROI — (Time saved per run × runs per week) − (authoring + maintenance cost).

Frequently asked questions

1.Is SDET a good career in 2026?
Yes — SDET is one of the highest-paid QA roles with average packages of ₹18–35 LPA in India and $120–180K in the US. Demand is rising with AI-augmented testing.
2.Do SDETs need to know DSA?
Basic DSA (arrays, strings, hashmaps, recursion) is enough for 90% of SDET interviews at product companies. Deep graph/DP is only asked at FAANG-tier companies.
3.Which language is best for SDET in 2026?
Java remains #1 in enterprise, Python for AI/ML orgs, TypeScript for modern web startups. Pick one and go deep — polyglot beats generalist here.
4.How do I switch from manual QA to SDET?
Learn one language + Selenium/Playwright + Git + one CI, then build 2–3 portfolio frameworks. Most switches take 4–8 months of consistent practice.

Was this article helpful?

Cluster · QA Career

More from QA Interview Questions

Behavioral, framework, coding — full interview prep.

Pillar guide · 11 articles
More in this cluster
From the QA Career pillar
Topic mapConcepts · Tools · People · Standards

Related concepts, tools & standards around SDET

A quick reference of the people, companies, frameworks and technologies most often mentioned alongside SDET in real QA teams — useful when you're mapping a learning path, preparing for interviews, or scoping a new project.

Core testing concepts
Design Patterns for Test FrameworksFlaky Test DiagnosisTest PyramidShift-Left TestingBehavior-Driven DevelopmentTest-Driven DevelopmentPage Object ModelContract TestingExploratory TestingRisk-Based Testing
Testing tools
EspressoXCUITestBrowserStack App LiveSeleniumPlaywrightCypressAppiumJMeterPostmanTestRail
Programming languages
JavaPythonJavaScriptTypeScriptC#SQL
DevOps & CI/CD
Certifications worth knowing
ISTQB Foundation LevelISTQB Advanced — Test AnalystISTQB Agile TesterCertified Selenium ProfessionalAWS Certified DevOps EngineerCertified ScrumMaster (CSM)
Companies hiring for this skill
GoogleMicrosoftAmazonMetaNetflixAtlassianThoughtWorksInfosysTCSWipro

Key takeaways

  • Master the fundamentals before tackling advanced SDET scenarios.
  • Always explain trade-offs — interviewers reward judgement, not memorisation.
  • Use real project examples; generic answers blend in.
  • Practice answers out loud — written prep doesn't transfer to live rounds.
  • Revise the 30-second cheat sheet the night before your interview.
  • Keep one strong scenario story ready for every section above.

SDET jobs hiring now

Live, indexable SDET openings — updated daily in Jobs Radar.

Browse all QA jobs on Jobs Radar

Loading current openings…

Home