Modern ML resume with Fortran – and it's from a 69K-subscriber YouTuber

1

I came across something interesting and wanted to share it with the group.

Egor Howell, a data scientist who also runs a YouTube channel with 69,000 followers, recently published a blog post on Towards Data Science about using Claude to build a “$200k+ ML resume.”

Here’s the blog: How Claude Help Me Build My $200k+ ML Resume
And his YouTube video on the same topic: If I Wanted to Become a Data Scientist in 2026, I’d Do This

What caught my eye: in his resume template, under Technical Skills, he splits languages into “Proficient” and “Experience” – and Fortran appears in the “Experience” category.

It’s genuinely refreshing to see Fortran getting a mention in modern data science/ML content, especially from someone with that kind of following. It clearly carries weight on a CV.

1 Like
2

In my perspective, this has been expected for a long time.

Because Fortran, from first day on, started his journey to look at the world through a mathematical lens.

In modern AI age, that way is much more impartant then never happened before.

Simply because, AI, in the base, is full of mathematical algorithms including different sub areas (linear algebra, probability … etc.) that has been focused learning.

Therefore, in my opinion, Fortran will be much more important day by day

3

Fortran naturally lends itself to Stats/ML/AI, and has a history in empirical modelling (e.g., Breiman 2001 introducing “Random Forests” with a Fortran implementation).

The s/w ecosystem gaps are closing (look at Neural-Fortran, FTorch, FStats, FastGPT, what I’m trying with FSML, some of Beliavsky’s projects, and many more) but

  • many great libs aren’t visible enough (the Fortran-Lang package list, Fortran Wiki’s lib list, Beliavsky’s lists, and other initiatives help, but many won’t be exposed to them),
  • many great libs aren’t easy to adopt for language newcomers; fpm is a game changer for how easy it becomes to use a lib in your project, but we’re missing accessible tutorials for a lot of great projects,
  • related to the above: those who are just starting to learn to code (e.g., university students) are often not even given the option to opt for Fortran despite its continued importance and institutions (like met services) having trouble finding qualified Fortran programmers (as was highlighted again at the last geoscience conference I attended).

So there’s a history, unused potential, but also some barriers to overcome here.

3 Likes
4

The fpm repository was intended to create a single stop for collections of libraries with long term availability provided to encourage use; and hopefully some vetting requirements to ensure documentation and unit tests and/or examples are included. That seems to have stalled out. Is that due to technical or legal or infrastructure and support issues or what? Does it seem like it is not worthwhile or won’t succeed? It seemed like the natural place for various approaches to compete for being part of a standard library, among other envisioned reasons it was needed.

2 Likes