🧪 QA Roadmap Manual · Automation · AI — an open curriculum by Arifuzzaman Antor

Stage 8 — The Path to Automation (without abandoning your craft)

Goal: know when automation makes sense, what to learn first, and how to make the transition from a position of strength — as someone who already knows what’s worth testing.

First, the correct mental model

Automation doesn’t replace testing — it replaces re-checking. A script verifies that known behavior still holds; a human explores, questions, and judges. The best automation engineers were good manual testers first, because a fast suite of the wrong checks is worthless. You already know what’s worth checking. That’s the hard part.

When to automate (and when not to)

Automate ✅ Keep manual ❌
Regression run every release Anything run once or twice
Smoke suite on every build Exploratory & usability testing
Data-driven repetition (same flow, 50 inputs) Rapidly changing UI (you’ll rewrite weekly)
API contract checks Visual “does this look right?” judgment
Precise timing/race conditions humans can’t hit Edge cases needing human setup judgment

ROI rule of thumb: value = (manual runs saved × time per run) − (build cost + maintenance). Maintenance is the silent killer — flaky suites get ignored, and an ignored suite is worse than none.

The learning path (in order — 3–6 months of evenings)

  1. API automation first, not UI. Postman collections you already have → add test scripts → run via Newman. APIs change less than UIs, so your first suite won’t demoralize you.
  2. One language: JavaScript/TypeScript or Python. Variables, functions, loops, objects/dicts — you need less than you fear. Free: freeCodeCamp JS or Python basics.
  3. Playwright (my recommendation, and what I run in production for client platforms) — modern, fast, auto-waits kill most flakiness that plagued Selenium. Their docs + codegen tool let you record-then-read real tests from day one. Learn Selenium concepts too — job listings still ask.
  4. Locators & the Page Object Model — the difference between a suite that survives UI changes and one that dies.
  5. Git + CI basics — push your tests, run them in GitHub Actions on schedule. “My tests run automatically every night” is a career-changing sentence.
  6. Reporting — Allure or the built-in HTML reporter; results nobody sees don’t exist.

Then, if the path pulls you: performance (k6), mobile (Appium), and AI-assisted testing (self-healing locators, AI test generation — this is where the industry is heading, and where I now spend my time via AZAI Labs).

Your first real project (the portfolio version)

Automate the top 10 regression cases from your manual portfolio project:

  1. A smoke suite in Playwright: login, core flow, checkout — running in GitHub Actions on every push
  2. An API collection with negative cases, running via Newman in the same pipeline
  3. README explaining what you chose to automate and why — that paragraph impresses more than the code

The titles as you go

Manual QA → QA Engineer (hybrid) → Automation Engineer / SDET → then everything opens up: Senior SDET, QA Architect, DevOps-adjacent reliability roles (my Mastercard chapter — chaos engineering — grew from exactly this path), or leadership.

The one warning

Don’t become the person who automates instead of testing. The market is full of flaky-suite babysitters who lost the testing instinct. Your manual foundation — risk thinking, test design, bug intuition — is the moat. Automation is leverage on top of it, not a replacement for it.


This chapter was the decision; the hands-on build is Stage 14: Automation Deep Dive — Playwright/Selenium, the pyramid, Page Object Model, locators & waits, data-driven design, CI/CD, and framework architecture.

Then the rest of the specialist tracks, where 2026 pay lives: AI × QA (AI agents, MCP + Playwright, testing AI systems), API Testing Deep Dive, Performance Testing, and Security Testing. Plus two chapters that matter as much as everything above: Soft Skills and the Sources & Further Learning library.

Star the repo, do the exercises, build the portfolio, and when you land the job or the promotion: tell me. I mean it.

← Back to the roadmap