Go for #2, but first spend a month or two seriously investigating potential startup ideas. Make it your job, 60 hours a week, to brainstorm ideas, research competition, create simple MVPs, etc. You likely haven't put as much effort into potential startup ideas as you have into your regular job, so don't feel inadequate or hopeless if you haven't stumbled upon the right idea yet.

In my experience, #1 never really works, especially if you are unsure of what the project will be in the first place. You have two years, which is more than enough time to build and test at least a half-dozen serious projects. If none of them work, go get a job with an awesome new portfolio.