I'm constantly learning, and seeing when things become "possible" with the current state of knowledge. Whenever I see that, on my own, I'll sketch out how the solution would work.

Then, I look around for companies that need a problem I already have a solution for solved, and I pitch my solution to them. Sometimes, I'll take jobs to solve problems I haven't considered before, but that's rarer.

If they like my solution, I set up a contract to implement it for them.

I'd say about 50% of the time I pitch a pre-existing solution, I get work. What's great is I'm essentially picking my own work and I get paid well (because the problems are hard).

The downside is you pretty much always have to be coming up with new solutions to things, and stay very current on technology.

I'm generally into something long before the "early adopters" get there, and I'm gone by the time it hits the mainstream. For example, I was working on a Google Spanner-like database while Google was doing the same.

At the time, no-one thought something like that was possible (outside of Google -- and they didn't tell anyone). I implemented different parts of the solution with three different clients over a two year period, none of whom hired me to create something like Google Spanner, but all of whom needed one part of the whole package solved.