Experienced devs have been through several loops of this process. Regardless of the theory behind each word, in practice it's just code words to prove you're part of the in group and updated.

In the past you might have been asked to add to the CV that you're into devops when you're just expected to trigger deploy jobs, or to explain that you take data-driven decisions when you're just adding some metrics to a powerpoint to justify a move that would have been done anyway, or to define yourself as a product engineer when they just want you to give a crap about users and be able to interface normally with managers.

Now there's a lot of "forward deployed engineers" that are actually just consultants or people interfacing with third parties, or AI-native devs that are just a guy using claude and maybe touching a couple mds.

This is not to say that significant changes aren't coming to the field, but the job application buzzword game is business as usual.

> Or pivoting somehow?

Existing companies are converting their desirable staff to these titles.

Half of my team are now "AI engineers" of some sort. It's really just indicative of who uses Claude/Codex the most.

It's not that different from when "sys admins" became "devops engineers", "db administrators" became "data engineers", etc.

From what I’ve seen jobs that need deeper domain knowledge will call out “ML” or “research” in the job title.

Everything else is just an inflated title to make the company seem AI forward.

Yes. It’s pretty trivial to pivot to, for these non-training/inference roles it’s just building systems where agents are the core implementation detail.