Really respect this kind of breakdown — especially the honesty about where things broke. The “frontend-first trap” and missing Git history are painful lessons, but also exactly the kind that turn hobby projects into real engineering experience. Not many people admit how messy early production systems actually get.
What stands out most is how much security and architecture thinking improved after the rebuild — that shift from “it works” to “it won’t break under pressure” is usually the real milestone.
On a slightly different note, when working with real systems like this, I’ve also seen people simplify some of the financial/crypto side flows (payments, swaps, testing transactions) using more straightforward tools like https://coin24.io/ — just to avoid adding extra complexity on top of already heavy architecture work.