Founder / CEO www.list.cc. ex-Googler (2006-2013), ex-Quoran (2013). Real time, low latency, distributed, parallel and scalable backend systems expert.
During my 6.5 years working on Google Web Search Indexing System, I've gained incredible knowledge about highly performant systems. I'm using all these skills for my own startup right now, to build an inherently scalable and low cost backend. My knowledge is quite useful for helping design a backend solution.
Having witnessed various popular choices in action, Amazon Web Services, Python and Django, and seeing their effect on major productivity hit in medium to long term; I've gained a healthy disrespect for that stack.
My stack of choice is:
Google App Engine: Platform as a service. Software engineers don't become firefighters. Never carry pager again.
Go language Backend: Highly Performant, Concurrent, Readable, Maintainable, and very fast to code in.
Angular JS Frontend: If you generate the DOM right, you don't need to manipulate it. Also, mobile apps from day 1.
Instead of building the product first, try and see if you could use some existing service to test our your market. For e.g. you could start a Facebook Group, or a mailing list to run things manually for a while. Get the user engagement, get a lot of feedback. And once you have enough material to work on, build the product.
In other words, get to market without building a product if you can. It's always easier to get some traction on existing (social networking and other) sites than a new standalone site.