Grigory and Oleg shows you can use TensorFlow based model within your Elixir application. They are not trying to recreate everything in Elixir. Rather to use existing tools, so training is done in Python (as you would expect). Interaction with model is achieved via Tensorflex which provides bindings over TensorFlow C library. Article also shows use of NIF for computation heavy (image resizing in this case).
Speaking of NIFs – very nice episode of Elixir Internals podcast with Dave Lucia of SimpleBet. They have quite nice ecosystem of Elixir offloading some of the heavy computing to Rust based NIFs via Rustler.
Article arguing that you should never use
:timer.sleep/1) in your tests.
An excellent write up on how to wrap existing C library with Rust to avoid unnecessary rewrites. While still getting a lot of Rust benefits. BTW if you need to do something similar with Python, you should check Building a Python C Extension Module.
Just interesting stuff
Nicely done video from Veritasium about strange instabilities of spinning objects. Known as as the Dzhanibekov Effect or Tennis Racket Theorem.