I'd start with some small, low-pressure, easily-completed writing/coding/design exercises. Speed-IF can be a good format for this, because it forces you to produce a tiny-but-playable piece to a deadline, going through a microcosm of the concept->design->implementation->test->release cycle, and you can release a game that a few people will play (and perhaps comment on) without the anxiety-inducing expectation that it'll be any good. There's one running
right now.
That particular format may not be your cup of tea, of course, but the best practice for producing games, generally speaking, is
producing games.If you think your strengths lie in a particular area, like writing, then you could plausibly focus on that, but pick up just enough coding and design to produce a smallish, tested-and-playable-but-not-wonderfully-robust, but well-written game; release it where it'll get some attention (
Introcomp might be a good venue for that) and then look for a coder. Your chances will be much better if you're a known quantity, and even if you don't find someone to work with, you'll have boosted your skills through practice.