prateek kukreja

Prepping for SDET roles :: topmate.io/prateek_kukrej…

15 yrs in SDET taught me the hard way: flaky tests aren't a Playwright problem, they're a waiting problem. Stop using fixed sleeps. Start using web-first assertions + auto-waiting locators. One change, half my flakes gone. What's your worst flaky-test horror story? 👇

Not sure where your test skills sit right now? I built a free 5-min SDET scorecard — covers performance testing, automation depth, CI/CD, and more. Link in first comment 👇

Most SDET bootcamps still teach JMeter in 2026. Most job descriptions I'm seeing now ask for k6. Here's the honest breakdown — and which one to actually learn: 🧵

Check your actual score: sdet-scorecard.vercel.app

TL;DR for 2026: • Add Playwright if you're Selenium-only • Docker + GitHub Actions > another framework • Call yourself an SDET, not a "QA Automation Engineer" (pay data backs this) Not sure where to start? Drop me a DM or book a session 👇

Playwright job postings tripled between 2024 and 2026. Selenium still leads on volume — but Playwright is growing faster, paying more, and dominating new product company hires. Here's what the SDET job market actually looks like right now 🧵

If you're at 3–6 yrs exp and the pay bump isn't coming — run this check: Is Docker on your CV? GitHub Actions? Can you wire up a Playwright suite in CI from scratch? If not, that's the ceiling. Not your seniority, not your framework choice.

The raw numbers (India, Feb 2026): • Naukri: 2,496 Playwright vacancies • LinkedIn India: 1,000+ Playwright roles • Selenium: still the most common, but growth is flat Volume tells you where legacy work is. Growth tells you where your career goes.

What 60%+ of SDET job posts in 2026 actually require beyond the framework: → CI/CD (GitHub Actions, Jenkins) → API testing (Postman / REST Assured) → Docker → SQL / DB testing → Git The framework is now table stakes. Infra skills are where the real gap is.

I've interviewed SDETs at Amazon / Visa / PlayStation /Paytm among others. Most fail for the same 7 reasons — and can't name their own weakest one. So I built a free 2-min SDET Readiness Scorecard: score + your 2 blind spots + next move. No sign-up. sdet-scorecard.vercel.app

Prepping for SDET roles :: topmate.io/prateek_kukrej…

Spent 15 years at Amazon, Visa, PlayStation, Paytm building test infrastructure. Coaching/mentoring space grew without me. Now I'm open on Topmate — SDET career calls, mock interviews, real advice from FAANG/FinTech SDET. Late to the party, but you'd rather have me here than not

Trick 7: Reset when the thread gets messy. Long conversations compound errors. A confused Claude gives worse output — not better. Start a fresh chat with a clean, tight prompt instead of continuing a derailed thread. Fresh context = sharper responses = less usage wasted.

Trick 6: Chunk big tasks into phases. Asking Claude to "build my entire landing page" = one massive context-burning task. Asking it to "write the hero section" then "now the features section" = same output, distributed, with space to course-correct between each step.

Hitting Claude's limit by lunch? It's your context window, not the model. 7 free tricks to stretch one plan ~5x 👇 #claude #claudeai