Couple of things: 1. Do you think you are in the right job, right now? The way you're thinking about the architecture rehaul and the business aspects of the project you should be the architect/CTO of the company :-) 2. If you are convinced the existing thing might not work, go tell your boss, sit with the team and make the changes. Starting out on your own is very time consuming, risky and tests your patience like anything, burns your enthusiasm out very quickly (speaking from experience) 3. I suggest fix the obvious gotchas in the current implementation within the current platform with minimum deviations and launch. In the meanwhile start a new effort, get a team, budget and targets. Work on your implementation, learn from the current deployment. Of course you may not be able to make millions out of this, but your risks too are limited.

But whatever you do, try not to leave it in the middle. Go the whole hog, enjoy the experience.