Interview questions about testing, answered one at a time.
I am preparing for an SDET-2 interview against a syllabus I built from a hundred and sixteen real job descriptions. It is fixed at 428 items across 13 topics, and I do not add to it mid-lesson.
A concept goes up here only after I have explained it back in my own words. Everything below is the map; 11 of them are written so far, and the rest are what I have not reached yet.
Crisp definitions + live test-design + metric literacy + own the CI quality gate & evidence-backed go/no-go. STOP before ISTQB cert, TMMi, company-wide QE org strategy.
Design/own/defend a reliable CI-friendly Playwright framework; Selenium literacy (52% of JDs) + BDD/Cucumber. STOP before self-healing engines, custom runners, grid infra.
Own REST/auth/JWT/schema/chaining + one strategy layer (contract+mocking+microservices pyramid). #1 cluster (69%). STOP before OAuth crypto internals, SOAP mastery, Kafka internals.
The 'design from scratch' narrative + core patterns + fixtures + CI/parallel. STOP before test platforms, custom runners. Over-patterning is a negative signal.
Full OOP/decorators/generators + deep Pytest; DSA ONLY Easy→comfortable-Medium (0/90 JDs ask Hard). STOP before LeetCode Hard/DP, metaclasses, CPython internals.
Application-level TS to read/write/maintain a Playwright suite (async/await is the real gap). Unlocks the 13% TS-only roles. STOP before compiler-level type gymnastics.
Own the TEST STAGE — 'I consume the platform, I own the test stage.' Jenkins slightly outranks GH Actions in our pool. STOP before Jenkins admin, K8s ops, Terraform, deployment mechanics.
Competent user of ONE tool (k6) with a CI gate + results literacy. SHOULD-level differentiator, not a MUST gate. STOP before LoadRunner, JVM tuning, capacity modelling.
Up to window functions + CTEs; SQL-as-test-oracle. A hard MUST (~35/50 JDs), not a quiet filter; ETL reconciliation is a MUST for Tekion/Cognizant. STOP before query tuning, DBA ops, pipeline authoring.
Fluent USER to run/debug tests, never owner. Docker highest ROI; K8s = actually-run kubectl (Apply). STOP before certs, Terraform/IaC, cluster ops.
Build & own an LLM eval harness in CI (Part A) + AI-assisted testing is now table-stakes (Part B). Capability-first vocabulary. STOP before training/fine-tuning, transformer internals, MLOps.