Julia sits kind of at the pinnacle of the numerics stack. It is a generalized very high performance language specifically designed to make mathematical and scientific computing clearer, more enjoyable, and more correct. In the techascent stack, it sits just after something like Smile's BLAS binding but just before TVM. Unlike a specific blas binding, you can write bespoke code in Julia and it is highly optimized for you. Unlike TVM, the language is Turing complete.
Julia has the best solvers known to computer science and it has a fully differentiable computing stack. I think it points the way forward in terms of dense and sparse style numeric computing. It has great metaprogramming capabilities and a community of extremely capable people pushing it forward.
Pairing Julia tightly with Clojure allows us to use their vast libraries of scientific computing and gives us an avenue to take advantage of truly cutting edge numeric research directly. I would like Julia people to enjoy using Clojure and vice versa.
This is absolutely a hair-on-fire early release. I put this forward in order to just let everyone know it is concretely possible and to open the door for people who may be interested in helping out.
https://github.com/cnuernber/libjulia-clj
Enjoy,
Chris