There are a lot of backend data science applications using Julia. I was involved in a 3D printing startup and we developed a path planner beginning with Julia 0.3. You generally have to precompile your dependencies into the base library, but once this is all containerized things run great (pretty much as fast as C++ in our case with a little bit of JIT overhead in the cornercases), and you just have to create some sort of API for the service. For backend, long-running, services it is a great language now. The client-side application/static compilation story is massively improving and is fortunately mostly possible outside of the core languages development. E.g. the infrastructure is there in the core language for static compilation, but the interfaces are still developing.
Do you have a guide on how to compile packages into base image?
There are some docs here: https://docs.julialang.org/en/v1/devdocs/sysimg/index.html
There is also now PackageCompiler that provides a nicer API without requiring you to build from source: https://github.com/JuliaLang/PackageCompiler.jl
Additionally, a lot of use cases were discussed at the recent JuliaCon (and in years past): https://www.youtube.com/user/JuliaLanguage/videos