Julia is fantastic if you do numerical work and want to write out your inner loops explicitly without sacrificing too much performance, either for pedagogical reasons or because you want to fiddle with the algorithm.
It is a lot of fun to start from an empty file, add maybe an import LinearAlgebra, and develop things like a convolutional neural network or a Markov-chain Monte Carlo algorithm completely from scratch. And then it is very rewarding to have such a program be fast enough (looking at Python here) to train on the MNIST dataset or find reasonable estimates for critical exponents.
I am not sure there are languages better suited than Julia for these kind of things.
Julia also has some of the nicest GPU abstraction frameworks of any language, achieving pretty low friction portability over a decent subset of CUDA, Metal, ROCm and CPU. KernelAbstractions.jl, AcceleratedKernels.jl etc.
It is a lot of fun to start from an empty file, add maybe an import LinearAlgebra, and develop things like a convolutional neural network or a Markov-chain Monte Carlo algorithm completely from scratch. And then it is very rewarding to have such a program be fast enough (looking at Python here) to train on the MNIST dataset or find reasonable estimates for critical exponents.
I am not sure there are languages better suited than Julia for these kind of things.