One recommendation, stolen from patio11: sell weekly blocks of your time. This is more fair to both those hiring you, and to your startup, because ramp-up and context-switch time are an expensive part of software development.
This is what I've started doing. Right now I'm consulting for a TechStars company doing some Akka stuff while I work on my own venture.
I'm single and 22 so doing this for a few months brings in enough to float me for ~year, not including the savings I had from the BigCo job I recently left.
The real bonus, IMO, comes from the fact that your former consulting clients are your most likely bet for job prospects if/when your venture fails, assuming they like your work of course.
Even if you have low living costs, you should still consider opportunity costs of not working. A consultant making $100k per year who takes two years off to do a failed startup has effectively lost $200k in income.