With that said: IMO you don’t. Not right away. You go with what is familiar, what can allow you to move fast, and what will at least generally conform to the platform you will be publishing to.
Why? Because you don’t yet even know if it is viable. You need a MVP to see if it has any legs in the first place, and then iterate quickly until something does “stick”.
Once you are getting your first paying customers (or any users at all), then you can start yak shaving at a tech stack that will be future proof and performative and scalable and magically never get in your way and whatever other need you think you might have.
But to obsess over the tech stack before writing even a single line of code will waste time and effort immensely. It’s not only like premature optimization, it actually is that implemented in the planning stages - how will you know what you need for your project before you build it? Projects have a way of throwing strange shit at you out of left field, and being agile and quick to adapt will help you more than any premature optimization in the tech stack.