Here's what I've learned from talking to many startups:
1- First step would be a casual phone call to learn about each other. Make sure you also learn about them and ask them tough questions about their business. Both of you are interviewing each other: it's not just that they're interviewing you. You don't want to suck up to a startup which you don't believe in and end up signing an offer there.
2- If all goes well, most will schedule a Skype coding interview. You just have to be prepared for those and that comes from experience (which you should have from your open source experience). Don't pull an all-nighter learning algorithms as they will almost certainly ask you a question that you couldn't cover the night before. Hints: learn a dynamic interpreted and use it (Python, Ruby etc...). I made the mistake of doing those in C in the beginning and got stuck on the details / writing more code that they wanted in an hour long interview. Some were impressed by my C skills, but that's not exactly what they're testing here. Instead, most are looking at the bigger picture at this point. i.e: they're seeing if you can you write a quick draft solution, and optimize it later instead of seeing if you can properly allocate a struct in C and place different struct types in one list. (Writing such an array takes quite sometime --> they might be interested in how you'd design it instead of actually doing it at this point).
3- Then comes the onsite interview. This is where you get to meet the team, see the office, the location and experience a day at that startup. This is very important. Your initial gut feeling is more important than you think. If you don't feel comfortable, then do not sign an offer. You'll be at that office at least 40 hours a week with the same people. You MUST be very comfortable working with them.
Startups will often interview you for at least a few hours on different areas. Design, coding, algorithms etc... Just do your best and the single thing that they care about here is that you can actually write code that works and make good decisions. It does not have to be perfect, but it has to work. They're looking for someone who can build something roughly in a few hours, and then explain the shortcomings (so that it can later be optimized). Someone who can get things done, and perhaps become a leader later on. Since you've contributed to open source, you should have absolutely no problem here.
Good luck!