Career tracks: SDET, quality engineer, AI quality
Where the roles are heading and what each actually asks for day to day. Honest about which are growing, which are consolidating, and what the AI-quality speciality currently pays for.
A tester with five years of experience recently asked me whether she should chase an SDET title or hold out for something in AI quality, and the honest answer was that the titles themselves tell you almost nothing. What matters is what a team actually expects you to produce day to day, and that varies enormously between two companies using the exact same job title.
SDET: still the biggest bucket, but consolidating
SDET roles are the most common QA career track, and they are quietly narrowing. Ten years ago an SDET might own manual test plans, write some Selenium, and file bugs. Now the job increasingly means owning CI reliability, writing framework code other testers use, and being the person who gets paged when the pipeline goes red at 2am.
This consolidation is real: companies are hiring fewer manual-only testers and fewer purely automation-writing testers, folding both into one role that expects both. If you are on this track, the highest-leverage skill is not a specific tool. It is being the person who can debug why a suite of 4,000 tests suddenly takes twice as long, which pulls directly from flakiness and CI pipeline knowledge.
Quality engineer: the title that means the most different things
Quality engineer is used by some companies as a rebrand of SDET and by others to mean something closer to a quality-focused generalist who influences requirements, coaches developers on testability, and owns quality metrics across a whole product area rather than one codebase. Read the actual job description, not the title, because the salary bands and daily reality differ sharply between these two versions.
A mid-sized fintech team I know hired three "quality engineers" in one year. One spent 80 percent of her time writing Playwright. One spent most of his time in requirements reviews and almost no time writing code. The third split evenly. Same title, same level, three different jobs. If a recruiter cannot describe what a typical Tuesday looks like for the role, that is a signal to ask more questions before accepting.
AI quality: real, growing, and still forming its own rules
The newest track, AI quality or LLM quality engineering, is genuinely different work: building golden datasets, writing evals, catching hallucination and groundedness failures, and reasoning about nondeterministic systems where the old pass/fail test model breaks down. Reading evals: the new test suite is a reasonable starting point if you are curious what the daily work looks like.
Pay for this speciality currently runs ahead of general QA roles at companies that are serious about it, mostly because so few people have real experience. But there is no established ladder yet: no clear junior to senior progression, no agreed-upon interview format, and a real risk that some AI-quality job postings are just QA roles with a trendier title attached. Vet these carefully by asking what specific eval or quality problem the team is trying to solve, not just whether AI is involved.
Building proof of work matters more here than in the other tracks precisely because there is no certificate or established interview loop to fall back on. A small public eval harness against a real chatbot demo tells a hiring manager more than a resume line claiming "AI testing experience."
- SDET: broad automation and CI ownership, consolidating with manual roles, most common and most stable demand.
- Quality engineer: title varies wildly by company, ask about the actual day-to-day before assuming scope.
- AI quality: newest, best paid relative to experience required, but immature hiring processes and no fixed ladder yet.
A junior tester named Marcus faced exactly this uncertainty two years ago. His company renamed his SDET role to "quality engineer" with no change in pay or scope, while a competitor across town used the same title for someone leading a whole quality practice. He asked both hiring managers for a week-by-week breakdown of the role before accepting an offer, and that single question saved him from a lateral move that would have looked like a promotion on paper only.
None of these tracks is strictly better. A tester who is genuinely energized by CI internals and framework design will be miserable forced into a requirements-heavy quality engineer role, and someone who dislikes ambiguity should be cautious about jumping into AI quality before the field has settled its own norms.
FAQ
Questions people ask
Is SDET a dying title?
No, but its scope is broadening. Most SDET roles now expect both automation authorship and CI or pipeline ownership rather than one or the other.
Should I specialize in AI quality without prior LLM experience?
You can, but build public proof of work first. The hiring bar is inconsistent and a demonstrated project carries more weight than in more established tracks.
How do I tell if a quality engineer role is actually an SDET role in disguise?
Ask what percentage of a typical week is spent writing automation code versus in requirements or process work. A vague answer is itself informative.
Does AI quality pay more than traditional QA roles?
Often yes, at companies genuinely investing in it, mostly because qualified candidates are scarce. That gap will likely narrow as the field matures.
Mark this complete once you have done the practice in the guide, not merely once you have scrolled to the end. The tick stays in this browser; nothing is sent anywhere.