n=430
AI-market corpus
measured 24 Jul 2026Read in full by a human, one posting at a time, coded against the 85-item taxonomy (60 technical skills).
Data provenance
Where every number comes from.
Every figure on this site carries an n= and a date. This page is the method behind them: what was collected, how it was read, and what the numbers can and cannot claim.
The measurements
3,000 postings, three corpora, two measurement days.
n=270
ML-engineer corpus
measured 24 Jul 2026Same human full-read method, same taxonomy, collected separately so the two markets never blend.
n=2,300
By-role sweep — five role families
measured 27 Jul 2026AI Engineer (498), GenAI Engineer (452), Agentic AI Engineer (510), Senior AI Engineer (441), LLM Engineer (399). Five parallel research agents fetched every full job description from public ATS APIs (Greenhouse, Ashby, Lever, Workable, Workday, SmartRecruiters) and the July 2026 HN hiring thread, then scanned each against a curated per-role lexicon (~95 terms). Full descriptions only, one count per skill per posting.
430 + 270 + 2,300 = 3,000 job postings read across July 2026.
Refresh cadence
Re-measured quarterly.
We re-measure the skills AI jobs ask for every quarter, so what the course teaches stays current with what the field is hiring for.