I mean, nothing in particular is stopping you. the IDE is the obvious bottleneck but someone this adamant can make VS code work.
Everything else is trivial. Fork AOSP for an environment to test in, find a non-Oracle Java to build the app for (if that fails, the NDK in c++ will work, but complicate matters), figure out the buildchain to produce your APK, and upload to a store like F-Droid.
Your limitations will come based on potential phone features you require. I imagine camera access or permissions may hit a few pitfalls, and I'm unsure in 2025 how easy those are to overcome (they were nearly impossible the last time I professionally developed in 2013). But Google open sources more than you'd think, so you have a lot of tools at your disposal.
-------
Now, is it practical? No, not really. You're playing on hard mode and will probably fall into a lot of non/underdocumented behavior. You're gonna be a trailblazer for all sorts of issues only the largest, most ambitious companies run into (and probably fix inhouse, so good luck researching others' solutions). Monetization opportunities will be shot, and any attempts to expand platforms will have you hit brick walls as Windows/Apple OS's require non-FOSS apis.
You would only be doing this either for a hell of an impressive portfolio piece, or for ideaological reasons. Otherwise you're shooting yourself in the foot.
