What I got wrong about the gap between manual testing and automation — and what the gap actually was.
In my entire career, no interviewer has asked me a single question about my degree.
I spent years being quiet about it anyway. Every job posting said "Computer Science or related field," and I read that line as a door I would have to sneak past. So I over-prepared for questions about my background that never came, and under-prepared for the ones that did.
I thought the gap was knowledge. It wasn't.
Manual testing had already taught me the expensive part. How software actually breaks. Which edge case will embarrass someone in production. Where real users get confused and click the wrong thing. Which change is risky even when the diff looks small. That judgment takes years to build and nobody hands it to you in a course.
What I was actually missing was a language to write it down in. Loops, async, locators, a framework. That part took months, not years. Hard, but finite — and crucially, the kind of hard that has a visible end.
Getting that backwards costs people time. If you believe you're starting from zero, you plan like a beginner: another course, more theory, waiting until you feel ready. If you understand you're already carrying the hard half, you plan like someone learning a tool. Those are very different ninety days.
It was a login flow on a demo site. There was no AI to ask back then — just documentation, forum answers, and stubbornness. It took embarrassingly long for what it was.
But it worked, and I ran it again, and it worked again. That small win gave me a confidence I still carry, and it's the reason the roadmap I wrote insists on writing a bad first test early rather than reading for another month. The confidence doesn't come from understanding. It comes from something running.
Today I test AI agents for a living, and I sit on the panel that interviews people for the role I once assumed I'd never qualify for.
From that seat: nobody's degree comes up. Not once, in any interview I have been part of. What comes up is whether you can talk about something you built and something you broke, in enough detail that I believe you were there.
That's the whole thing. Depth of a real story beats breadth of a claimed one, every time. A candidate with one project they can walk through end to end — why that locator strategy, what broke in CI, how they found the cause — interviews better than a candidate with five technologies listed and nothing to say about any of them.
Which is also why the degree never comes up. It doesn't predict either of those things.
Three things I'd say to anyone making this move without a CS background:
Mechanical Engineering turned out to be fine preparation for this, incidentally. Tolerances, failure analysis, root cause. Different materials, same instinct.
If you're planning this move, the 90-day roadmap is the version of this advice with dates attached — what to learn, what to skip, and what to have to show at the end.