I reviewed 500 AI job postings last month.

The market has fundamentally shifted from what it was even 12 months ago.

What's disappearing: "Data Scientist" roles that are purely notebook-based analysis. Pure prompt engineering positions. Generic "AI/ML Engineer" roles with vague descriptions.

What's exploding: AI Infrastructure Engineer (building the platforms everyone else uses). LLM Application Engineer (production-grade GenAI systems). AI Safety & Evaluation Engineer (testing, red-teaming, guardrails). ML Platform Engineer (internal tooling for ML teams). AI Agent Developer (building autonomous workflows).

The most surprising trend: companies are hiring fewer ML researchers and more ML engineers. The research-to-production ratio has flipped. In 2023, it was 3 researchers per 1 engineer. Now it's 1 researcher per 4 engineers.

Why? The models are good enough. The bottleneck moved from "can we build a good model?" to "can we deploy it reliably, cheaply, and safely at scale?"

If you're job hunting in AI right now, optimize for deployment skills, not research skills. The market is screaming for people who can ship, not people who can publish.

Adapt your skills to where the demand is moving. Not where it was.