If your hypothetical project, and/or its components have any similar counterparts in the real world I would start by using those as the basis for your research. If not, and it’s 100% unique, then I would treat it as if it were a brand new product and research your why, competitor, demographic, personas etc as if you were going to legitimately build it. Even better, find real people to get feedback on your ideas/prototypes.
Treating it as a real product, and where possible using real data, will give you a far better result than if you make assumptions. Without real insights, you’ll have hollow results that I’m betting will be obvious in an interview/portfolio scenario.
Maybe create something that resolves a problem that you/your social circle genuinely have? It will make identification of your vision easier and the design criteria and Success Metrics should fall out of that.