How to become a peer-reviewer?
As an editor, when I need peer-reviewers, if I don't know someone with expertise in the topic at hand, I look to see who has published peer-reviewed work in a similar area recently. No-one signs up to do reviews, it's up to me to go out and find them. Which means, if you haven't published peer-reviewed work recently, you aren't likely to get any invitations.
So, in my case, if you want to do peer-reviews, the first step is publishing something in a peer-reviewed journal. Otherwise, no one will know you're there.
This doesn't mean someone who isn't publishing is never qualified to provide a peer review. But if I don't know you, and you haven't published anything yourself, I have no way to evaluate if you are qualified to provide a good review.
Another thing to keep in mind is that the distinction between an interesting paper, and an interesting publishable paper, may not be clear to someone who isn't actively pursuing academic work in the area.
In addition to Thomas’ answer, peer review (in my experience) is somewhat of an “invitation only” closed club: journal editors need to know your name, and they do so because you have published in the relevant (sub-)field. And even then, editors tend to favour senior researchers or people they otherwise trust.
Case in point, I’ve published several papers during my PhD and (co-authored) during my postdoc. Yet the only peer review invitations I have received so far have been due to a specific recommendation from former group leaders to the editors.
This will probably vary across fields. But the fundamental principle is the same: editors need to put some level of trust into the competence of the reviewer. And since the research happens at the cutting edge (otherwise it wouldn’t be very interesting research), the reviewers are often chosen from a fairly small pool of people known to work at that cutting edge. Unless you have recently done outstanding, publicly known work in a field of active research, it’s unlikely that an editor would consider you as a reviewer. Merely having proficiency in already established methods is not useful.
In general, and I accept that there may be some exceptional sub-field I do not know about, peer reviews in Computer Science are conducted by researchers. You need experience doing research (viz. writing papers) in order to understand the quality of what you are reading. Most people are introduced to peer review during their PhD studentship when a program committee member delegates one or more of their papers to a PhD student, typically their own student. Once someone has a PhD, they may then be invited as a program committee member or a journal reviewer by the program chair/journal editor.