Unlike Python, which I often use to implement industry-standard stuff, I often use Julia to try to push the limit of what’s possible or do something new.
And they often failed.
Too cursed to use.
Too slow. Slower than a web-based optimizer.
What should I do? I’m afraid I would fail again and produce yet another cursed package or something unusable or way worse than someone else’s yet again.
No matter where I look, I see amazing things, and no matter where I look, I don’t see a crack where I could win.
Nobody writes amazing things as their first major programming project.
Learn to program by writing simple stuff that is fun, without worrying whether it is world-shaking or even especially “good” code. Learn to work on bigger projects by contributing to someone else’s bigger project in a small way (e.g. fix a bug or write some documentation or a test case). Then move on to contributing in a bigger way (add a feature to someone else’s package, or write a package that does one useful small thing well, or port a small useful package from another language).
But I’m not the first person to tell you these things — you’ve made similar lamentations several times on this forum, and many people have given you the same advice. I’m not sure what more you’re looking for, but maybe it’s something that is better found offline (e.g. from a friend, a colleague, a teacher, a family member, …).