Key findings
- 96 public postings are classed as AI/LLM evaluation QA (2.5%); 149 postings have an AI, LLM or ML term in the title.
- 50% of evaluation postings are remote, against 22% for all QA postings.
- AI and ML companies account for 26% of all 3,863 postings (1,007), the largest industry in the sample.
- US hourly ranges in AI-titled postings (n=4) have a median midpoint of $62.25 an hour and run from $6 to $85.
01Two kinds of AI work
'AI evaluation' covers two kinds of work that employers describe with different titles. We separate them so the numbers are not mixed.
| Group | Postings | Share of 3,863 |
|---|---|---|
| Classed as AI/LLM evaluation QA | 96 | 2.5% |
| Title mentions AI, LLM, ML, model evaluation, annotation or trainer | 149 | 3.9% |
| Employer industry is AI and ML | 1,007 | 26.1% |
The first group is rating and testing model output: evaluators, annotators and trainers. The second adds engineers who test AI products or use AI tools in testing; 31 of the title matches sit in QA Automation Engineer or SDET roles.
02What the postings look like
The 96 evaluation postings by employment type, level and tools:
| Measure | Value |
|---|---|
| Permanent / contract | 88 / 8 |
| Junior / mid / senior | 7 / 55 / 29 |
| Remote / hybrid / on-site | 50% / 14% / 36% |
| Top countries | United States (32), India (15), United Kingdom (6), Poland (5) |
The most-named tools are Python, Postman and Playwright. In the Flex version of the work, you read model output critically, apply a rubric and sometimes script the evaluation.
03Skills and Flex tasks
Skills for the role as we describe it:
- Prompting
- Evaluation design
- Python
- Domain expertise
- Annotation tools
For Flex tasks, the AI evaluation band is $25-70 an hour over 2-8 weeks at 3-4 h/day. Example: Rate 1,500 assistant answers against a safety rubric. 2 AI evaluation tasks are open on the board now, at $1,650 for about 55 hours and $45-62 an hour.
04Pay
Pay depends on the kind of work, and public pay data is thin for both kinds. The Flex band and our model give reference points.
| Group | n | Median midpoint per hour | Lowest to highest |
|---|---|---|---|
| AI-titled postings with a US hourly range | 4 | $62.25 | $6-85 |
| Source | Rate or level |
|---|---|
| Flex AI evaluation band | $25-70 per hour |
| Our US model, mid-level AI/LLM evaluation QA | $104k per year |
Browse the AI/LLM Evaluation QA role, its salary guide and Flex AI evaluation tasks.
05Method and limits
Sample: 3,863 public postings harvested 20 Oct 2026 from employer career pages and public job feeds. Listings are postings, not people or confirmed openings: one role open in several cities counts several times (2,417 distinct titles across 3,863 postings). PassRidge talent-pipeline roles (149) are excluded from every statistic.
AI terms are matched in the title (AI, LLM, ML, machine learning, model evaluation, annotation, trainer). Postings that describe AI work in the body only are missed.
Pay uses ranges the employer published, converted at the dates and rules in our pay article.
The method for roles, tools and pay is documented on our salary methodology page. Spot an error? Write to [email protected].
Go deeper
Sources
- PassRidge public listing snapshot, 20 Oct 2026
- PassRidge Flex rate card