I'll answer this for the OP as someone that has worked on game engines for over a decade. One reason is that Unreal and Unity are awful engines. Bad code, bad performance, bad workflows, bad design. Having worked with both, and needed to customize/modify both heavily, it's amazing how bad code can get if the provider of the code isn't dogfooding their own stuff. Unity doesn't ship a game at all, and if you rely on an Unreal feature that isn't needed by Fortnite, good luck.
As much as all that may be true by various definitions of "bad" it's still a lot easier to trust Unreal or Unity, which thousands of people have shipped games with, than some homebrew engine that nobody's heard of.
You need to pick your poison. Are you going to battle Unreal quirks and bugs but at least get portability, or are you going to battle OpenGL, Vulcan, Metal and DirectX directly? These are pretty "bad", too, because their implementations vary so much from one OS to another, even one version of one OS to another. Then you need to deal with audio, game input, and dozens of other things that are equally quirky and broken.
I've known a number of people that have shipped with Unity, Unreal and Game Maker Studio. Most people that embark on building their own engine never ship a game because it's a very deep rabbit hole to fall into and one with no bottom.
It's really a matter of how "indie" you want to be. Overgrowth managed to ship with a custom engine despite the odds, but wow, was that ever an undertaking.
I've been pretty involved in just about every layer of the stack. I cut my teeth on graphics (and still do graphics) but I've had to deal with networking, input, audio, and gameplay/scripting code as well.
Platform support is likely the most expensive to do right, but I would argue supporting say 2*n engineers per platform is actually cheaper than needing to rewrite huge chunks of Unreal or Unity. I can honestly say with confidence that no AAA game has shipped on Unreal without needing substantial rewrite of various internals. I've consulted on several Unity projects, and it's an actual dumpster fire :/. I've heard that Unity's valuation is driven by cloud service offerings and the ad network they've spun up. It shows in the quality of the code and the investment. That said, frameworks like SDL2 have come a long way in democratizing the platform-dependent bits of code that used to be insanely hard to get right.
I really do believe a better option "should" exist for developers that want to make the game and not the engine, but I sort of lean towards building an engine (leveraging open source frameworks wherever possible) for more "serious" studios that need to maintain their game for several iterations.
It's "Vulkan" btw :) Also grats on shipping!
We're a place where coders share, stay up-to-date and grow their careers.
We strive for transparency and don't collect excess data.