I scraped 1,000 ML/AI engineer job postings

Summer is a dry season if you want to land a good job in tech. But, that doesn’t mean you shouldn’t prepare for Aug-September job openings, as the market starts accelerating again with good offers, projects and opportunities.

And even without the dry season market has been really vicious. All the 7 rounds of interviews, hours of tailoring resume, cover letter and portfolio just to be ghosted. Being a freelancer or a founder gives much more job security than corporate, and somehow no one prepares you for that.

That’s why I researched what’s in demand, using the good old technique of data scraping (these pay really well on the freelance market btw)

Here’s what six months of ML and AI engineering hiring data actually says, and why most candidates are still building resumes for a job market that stopped existing about 2 years ago.

This isn’t news, but the size of the number is the point. More than half of every ML/AI engineering posting names Python specifically, not “a scripting language,” not “your preferred stack.” Python.

If your strongest language is something else, you’re not disqualified. But you’re spending your first interview answer explaining a gap instead of demonstrating a strength. That’s a bad trade in a market this competitive.

This is the stat that should worry people who only know one framework. The gap between PyTorch and TensorFlow is small enough that it’s not really a framework war anymore. It’s a signal that teams switch stacks, inherit legacy code in the other one, or get acquired by someone who standardized differently.

The risk isn’t picking the wrong framework. It’s picking only one and assuming that’s a permanent decision. Engineers who can read both survive stack changes. Engineers who can’t get quietly filtered out of roles that never even mention the framework they’re missing, because the posting assumes you already cover both.