ML/AIWork

ML/AI Work Research · Demand intelligence

Demand × Supply gap: chased vs overlooked AI roles

Job seekers searching for “prompt engineer” outnumber live postings 555 to 1, while generative ai engineer draws just 4.0 — a 138× less crowded door into the field — the most chased and least crowded corners of the AI job market (ML/AI Work, search demand × live corpus, as of 2026-06-11).

Searches per live postingbar length is log-scaled · label shows the real ratio

Prompt Engineer
555
LLM Engineer
105
AI Agent Developer
72
NLP Engineer
71
AI Consultant
63
AI Product Manager
62
AI Infrastructure Engineer
35
AI Trainer
29
Computer Vision Engineer
24
AI Agent Engineer
21
MLOps Engineer
20
AI Engineer
17
Agentic AI Engineer
8.7
Machine Learning Engineer
7.5
AI Research Engineer
4.4
Generative AI Engineer
4.0

red = overheated (seekers ≫ jobs) · green = supply-rich (more postings than searchers — seeker leverage)

Prompt Engineer: 62,690 searches/mo against 113 live postings (555 s/p). At the other end, generative ai engineer shows 4.0 searches per posting — a 138× gap in seeker competition between the two ends of the chart.

Anti-hype: “prompt engineer”, global searches/month

prompt engineer (3-mo avg)
0K51K102K153Kprompt engineer (3-mo avg) — latest: 63KJun 22Feb 23Oct 23Jun 24Feb 25Oct 25Apr 26

“Prompt engineer” searches peaked in 2023-04 at 168K/month and have fallen 65% since — yet it still tops the gap chart above: interest decays slower than the role's actual market.

Methodology

Demand = average Google searches per month over the last 3 closed months (via DataForSEO), summed across 18 markets. Supply = live postings in the ML/AI Work corpus (active, AI-tier) whose normalized title contains the role phrase on a word boundary. The ratio is a direction-of-imbalance signal, not a vacancy forecast: search interest includes the curious as well as applicants, and title matching undercounts roles advertised under adjacent names. Corpus definitions: methodology.

Cite this

ML/AI Work, "ML/AI Work Demand × Supply gap", dataset v2026.06, n=16, as of 2026-06-11. https://mlai.work/research/demand-supply-gap