Quality Assurance

My vision for QA

Quality Assurance is an engineering gateway. A test engineer is a software engineer who specialises in quality, not “just a manual tester.”

Where quality comes from

A modern system is more than simplicity, intuitive design, and a solid backend. Those matter. From a QA lens you also need black box, white box, and grey box testing, security testing, CI/CD and SDLC discipline, test design, system stability, and quality by design.

You still have to live in a system before the cracks show. AI can speed discovery up. Used badly, it can also burn years of careful quality work.

Layers of assurance

Black boxWhat the user sees
Grey boxAPIs and integration
White boxCode and units
SecurityAbuse and data safety
StabilityLoad, soak, recovery
CI/CD gateEvery change, every time
Layers of assurance, outside in, then into the release pipeline.

AI in QA, if applied correctly

AI opens real opportunity for Quality Assurance. It also raises the stakes. Incorrect AI can destroy years of work. That is where QA comes in.

QA Engineering is, by design, a V&V guarantor. It checks that AI does what it is supposed to do, finds the flaws, and improves the system hand in hand with the builders.

Near-exhaustive coverage at scale

Truly exhaustive testing is still impossible. What changed is scale.

Automation engineers, SDETs, and correctly trained technical analysts can use QA swarms and local investigation to approach near-exhaustive coverage on the paths that matter. Breadth from swarms. Depth from humans.

Human and machine

HumanJudgement, context and the final call
MachineScale, speed and patterns at volume
Human improves machine. Machine improves human.

Human and machine improve each other

  • Human → machine: sets the bar, owns risk, writes the questions models miss, decides what “good” means.
  • Machine → human: covers more ground, spots patterns at speed, frees people for deep investigation.
  • Together: quality by design across Design, Build, Test, Release, Learn.

The perfect QA process in an AI world

  1. Design for testability, security, and clear contracts.
  2. Build with CI/CD gates, not hope.
  3. Test across black, grey, and white box, plus security and stability.
  4. Release only when V&V says the change earned trust.
  5. Learn from every failure and feed it back into humans and models.

AI assists every step. Humans keep the bar. That is Quality Assurance.

Quality by design loop

Design Build Test Release Learn AI assists
AI in the middle, humans keep the bar.

Swarm plus local investigation

  • Swarm probes for breadth
  • Human deep dive
  • Defect found
Swarms for breadth. Local investigation for the cracks that matter.

Why this matters to me

I grew up with hardware and software from the 1980s onward. I know thick handbooks and modern systems that feel obvious on first touch. I want fast and right in the same release. Skip the details and the user feels it.

Quality is more than finding bugs. Good systems feel trustworthy. In an AI world, QA is how we keep that trust.