What are companies actually asking in SDET interviews right now?

Interviewed at 12 places over the last year (some hiring me, some I helped hire). Here's the current mix at typical mid–senior SDET loops: **Round 1 — Recruiter screen (30 min).** Salary, notice period, "why leaving". Not testing you technically. **Round 2 — Hiring manager (45 min).** Behavioral + high-level: "walk me through a flaky test you fixed", "how do you decide what to automate", "tell me about a time you disagreed with a dev". *This round eliminates more candidates than the technical rounds.* Have 5–6 STAR stories ready. **Round 3 — Coding (60 min).** Two flavors depending on company: - **Product/SaaS shops:** LeetCode-easy or medium. Arrays, strings, hash maps. Not asked to invert a binary tree. Preferred language. - **Fintech / large enterprise:** language-specific fluency (Java collections, Python idioms), plus one "write a small utility" — parse a log file, deduplicate, group by. Test-writing might be part of the same session. **Round 4 — Framework / test design (60 min).** The most SDET-specific round: - "Design a test framework for {a new mobile banking app}. What layers, what tools, how do you organize tests?" - "Here's a flaky test. Debug live." (They share screen with a broken Playwright spec.) - "How would you test a URL shortener?" (Classic — hit them with functional, non-functional, security, edge cases.) **Round 5 — System design (60 min, senior+).** Not "design Twitter" — more like: - "Design a test data management system for 200 microservices." - "How would you build a CI pipeline that gives feedback in under 10 minutes for a monorepo?" - "Design a flaky-test detection service." **Round 6 — Bar raiser / cross-team (45 min).** Behavioral + culture. Same STAR stories, different angle. **What surprised me in 2025/2026:** - Way more emphasis on **CI/CD and observability** than framework choice. Nobody asked "Selenium vs Playwright?" — they asked "how do you know if your suite is healthy?" - **AI in testing** comes up in ~80% of loops. Not "will AI replace you?" but "how are you using it today, and where does it break?". Have honest answers. - Contract testing (Pact) mentioned at every fintech. - Zero XPath trivia. Nobody cared about `ExpectedConditions` chapter and verse. - Live debugging is more common than "explain retries in theory". Prep priorities: STAR stories > framework design > coding > CI/observability > system design > AI usage.