Step one is going to be to learn what NetFlow is -- in particular, NetFlow itself is a push technology ("NetFlow" is the name of the protocol over which information is sent; a UDP stream directed at a collector) rather than a pull technology.
It doesn't really make sense, then, to say that NetFlow is "installed"; typically one says that a Cisco device is "configured to send NetFlow". Perhaps they mean that a particular NetFlow collector is installed? NetFlow Traffic Analyser is a collector: it has a component that receives, parses, and stores NetFlow data for later retrieval. Showing only data from only one IP is a more difficult task, not a less difficult task, because it requires having all the data for any IP address that might legitimately ask, and being able to retrieve and format it on-demand.
So, my push in the right direction is that you need to find out: do you need to write a collector, or are you writing an interface to an existing collector?
Writing a collector is not easy, by the way, and is not typically a task for a web developer. If that turns out to be what you've been asked to do, I would highly recommend using an existing one rather than writing your own, because there are a number of pitfalls. There are a number of existing NetFlow collectors that have APIs and command line tools that can be used as the basis for an application like you describe.