Next two mystery projects after Nx, what could they be?

Jose announced the Nx project(s) today on the ThinkingElixir Podcast (#034 José Valim reveals Project Nx - Thinking Elixir). The podcast episode would have been the lead to the next steps in learning about Nx, and yet Jose threw a curveball at the end of the episode. At 1:15:03 of the episode he announces two new mystery projects. He says:

We talked about those three fronts, but we are already working on two other projects that are related to all this.


This is the foundation so what can you do on top of this foundation? We already have been exploring and considering those things…

Given what was unveiled in the episode. What do you think is in the works?

Since I’m not a Data Science/ML person I’m not sure what makes sense given the announcements and the sheer volume of work that’s left in delivering. If I had to guess it would be the Elixir Nx version of Keras (GitHub - keras-team/keras: Deep Learning for humans), Pytorch (GitHub - pytorch/pytorch: Tensors and Dynamic neural networks in Python with strong GPU acceleration), or SciKit (GitHub - scikit-learn/scikit-learn: scikit-learn: machine learning in Python).

Anyone have any ideas on what may be in the works? What makes sense?


Ny and Nz I assume…


Perhaps José will drop some hints on next Wednesday’s Lambda days.

We can deduce the 2 other mystery projects based on the conventions already on the Python side for Machine Learning.

Essentially if Nx is the Elixir equivalent of Numpy (which is modeled after Matlab) and Jax (which is Numpy on steroids). In fact, José mentions Nx as the foundation. This is similar convention in the Python world: where Numpy is the foundation and infrastructure for other machine learning libraries like Pandas, Matplotlib, SciPy and Scikit-learn ( which in turn is built on top of Numpy and SciPy).

Basically if Nx = Numpy, then the other 2 are most likely the Elixir equivalents to Pandas, Matplotlib, or SciPy or some combination of the three.

If I were a betting man, I would guess Pandas and Matplotlib. Although SciPy would be the real game-changer for me personally!


I wonder if it visualisations are one of the things will it more like ggplot2 than Matplotlib. Wondering because ggplot2 already has a very functional interface that uses pipes in R.