ac_alejos

TLDR: The problem domain (telecom) fits Elixir perfectly

If you’re talking about scaling this, Elixir is built on the BEAM VM which was originally made for Ericsson and is tailor made for telecom systems.

Its whole paradigm is built around the concept of Let It Fail, which is basically about achieving fault tolerance through isolation and supervision.

So aside from the fact that Elixir+Phoenix is a productive framework that allowed the author to build this in a few days, it also means that it will scale very well with minimal code changes.

For reference, one of the solutions you might use to distribute this in Python is Celery, which is built on RabbitMQ which is built on Erlang, which is the predecessor of Elixir.

Absolutely! The whole ML ecosystem has grown to where it is now in less than 3 years or so. It helps that theres a good amount of concerted effort behind it.

XGBoost is an extremely popular and powerful machine learning library that implemented gradient-boosted decision tree models, and is still state of the art in field such as time-series analysis and tabular data prediction while taking a fraction of the computing resources of models such as neural networks.

Elixir now supports XGBoost models (even those trained outside of Elixir) and you can leverage the benefits of Elixir concurrency and distribution to serve your decision tree models using Elixir-Nx's `Nx.Serving` construct.

In this blog post (which you can run as a Livebook), we walk through a full example of training a spam detection model using XGBoost and serving it with Elixir. Check it out!