Downside of contract work is that other people don't pay for your training. A good company will often get you some damn good training and teach you excellent basic training. Especially if it's the same field. I probably learned more tricks in probation time than I did in all my time making games by myself, and the pressure helps you to learn faster.
But if a steady job is not your thing, you might want to consider professional part-time work. oDesk is a great place for freelancing. They keep track of what you do, so if you're doing a lot of decent work, you'd get a better reputation, and get some good paying stuff. So you can just do contracts for 6 months, until you get enough money to feed yourself for a couple years, then just quit and go indie.
Also, be wary of contracts. I could probably write a book on how they can go wrong (normally with little cheating on their end). But some simple things.. make sure you lay out and document
exactly what they want. Charge for extra features, or you'll spend months on polish. And make it crystal clear that mobile layouts don't end up like the sketches do.
My roadmap is: to make small games for a while, learning by failing and finishing games. That way, I get more used to the workflow and tools, I get better at everything, I keep writing down ideas and working on them in my head, etc. Later on, I would try to make something bigger, and if I ever have something promising, then I would consider leaving my job.
By roadmap, I mean have the development schedule for your game clear. Every feature should be outlined, or else you'll fall into the feature creep trap. If you want to survive on donations/kickstarter, people should know that you're giving them, when to expect it done, how far the progress is exactly. If you're surviving on your own money, you still want a clear scope so you know how much time to waste on small features, and when to cut things out.
Doesn't have to be detailed, just spend a day, or even just a couple hours planning all the components along the way.
Don't make a couple. Make one really, really good one. Seriously - all the mobile markets are more than saturated right now. I'm not an expert on mobile, but it seems to me that unless you hit a mobile game right out of the park it will make pocket money at best.
Er, just my experience with touchphones - I'd say quantity beats quality. It's like learning programming over from scratch, because everything is designed differently. You will mess up your first few apps.. heck, I'm on my 5th and I still don't have a consistent design that I'm happy with. Probably one of the biggest traps is applying a lot of experience with PC development, especially typical assumptions on UI and concurrency.
You don't want to churn out lots of things of the same type, but you want to try lots of small, different things and learn from the experience. Something like making your first PC game - it sucked, and never became your big hit, but you learned from it.