Key findings
- 1
Among the expert fields, data and AI work has the most people per open role (18.1 members for each of the 68 roles naming its professions), and science, math and engineering the fewest (1 for each of 644): there the roles are short of applicants.
- 2
Writing & Content roles have the most members who strongly fit them (17.77 per open role), and Robotics & Physical-AI Data roles the fewest (0.79).
- 3
The most members per open role are in India (0.72), the United States (0.31) and Brazil (0.26); in the United States, 883 members share 2,894 roles open to them (0.31 per role).
Figure 1
People per open role, by expert field
In which expert fields are there more people looking for AI training work than roles open, and in which are roles short of people?
| Field | Members per open role | Members | Roles open on 26 Oct 2026 |
|---|---|---|---|
| Data and AI work | 18.1 | 1,231 | 68 |
| Business and management | 3.9 | 587 | 150 |
| Health and medicine | 3.3 | 733 | 220 |
| Software and IT | 3.2 | 1,263 | 390 |
| Finance and accounting | 1.6 | 408 | 255 |
| Law | 1.2 | 172 | 146 |
| Science, math and engineering | 1.0 | 635 | 644 |
| Every member and every open role | 1.8 | 7,850 | 4,295 |
Sample. 7,850 people with a profile by the end of 26 Oct 2026; the 4,295 roles open on 26 Oct 2026, 1,784 of them naming a profession of an expert field in their title.
A member's field is the field of the best fit their profile reading names. A role is in an expert field when its title names one of the field's professions, by the rule the profession pages list roles with (the expert fields: health, law, finance, business, software, data and AI, science and engineering). Data and AI work counts members whose best fit is data or AI work of any kind, annotation included, and the roles that name a data science or machine learning profession.
Language, writing, teaching and annotation roles mostly name no profession in their title, so their people and roles are compared by kind of work (the next figure). The last row compares every member with every open role.
No row with any open roles is a small sample: the rows differ so much that each reads on its own from its first.
No count below 3 people is shown, nor any share that would reveal one, so no figure can point at a person; a group too small to show on its own is counted in "Other".
Source. AITraining.jobs Research, AI Training Talent: Supply and Demand, live data.
Figure 2
Qualified people per open role, by kind of work
For each kind of AI training work, how many people who strongly fit it are there for every role open?
| Kind of work | Strong fits per open role | Members with a strong fit | Roles open on 26 Oct 2026 |
|---|---|---|---|
| Writing & Content | 17.77 | 622 | 35 |
| Data Annotation & Labeling | 13.49 | 3,223 | 239 |
| AI Evaluation & Rating (RLHF) | 7.84 | 1,215 | 155 |
| Coding & Software Evaluation | 3.79 | 1,229 | 324 |
| Transcription, Audio & Speech | 3.78 | 2,289 | 606 |
| Multilingual & Localization | 2.95 | 2,722 | 923 |
| Agentic Tasks & RL Environments | 2.41 | 190 | 79 |
| Red-Teaming & AI Safety | 2.40 | 125 | 52 |
| Law, Medicine & Finance | 1.73 | 2,658 | 1,537 |
| Math & STEM | 1.45 | 357 | 246 |
| Research Studies & Surveys | 0.90 | 19 | 21 |
| Robotics & Physical-AI Data | 0.79 | 62 | 78 |
| Every kind | 1.53 | 6,577 | 4,295 |
Sample. Members with a fit check or a role opened 29 Sep 2026 to 26 Oct 2026, against the 4,295 roles open on 26 Oct 2026.
Members with a strong fit: the members the judge found a strong fit for at least one role of the kind in the window (one member can count for several kinds). Checks are made for the roles the site lists for each member and the roles members ask about.
Members who opened one: the members who opened a role of the kind on the site in the window.
No row with any open roles is a small sample: the rows differ so much that each reads on its own from its first.
No count below 3 people is shown, nor any share that would reveal one, so no figure can point at a person; a group too small to show on its own is counted in "Other".
Source. AITraining.jobs Research, AI Training Talent: Supply and Demand, live data.
Figure 3
People per open role, by country
In which countries do the most people compete for each AI training role open to them?
Sample. 7,850 people with a profile by the end of 26 Oct 2026, in countries with at least 3; the roles open on 26 Oct 2026 that a resident of each may apply to.
A role is open to a country's residents when it is open worldwide, states no rule, its list names the country, or its exclusion list does not (the same rule as the job board's country filter).
Job-alert subscribers are the people subscribed to the site's job alerts from the country, most of them members too, so the two columns overlap.
Rows with fewer than 12.2 open roles are marked as a small sample and not ranked: below that the average of the rows weighs more than a row's own open roles (the weight fitted on the rows' own counts).
No count below 3 people is shown, nor any share that would reveal one, so no figure can point at a person; a group too small to show on its own is counted in "Other".
Source. AITraining.jobs Research, AI Training Talent: Supply and Demand, live data.
For labs, data vendors and researchers
Buy the data behind this report
The figures above are a summary. Behind them are records we collect ourselves, every day:
11,094 roles tracked, each with its history
Every role on 56 job boards, read daily since 13 Jul 2026: its pay and place, and the first and last day it was open.
37,580 apply clicks in 28 days
Which roles people actually apply to, by field, pay and kind of work. It happens on this site, so nobody else can count it.
39 lab and vendor hiring boards
The job boards of 25 AI labs and 14 human-data vendors, read every day: who is hiring for human data, and who is hiring to buy it.
What you can buy
The records
The rows behind any figure, with their history, not only the totals.
A regular feed
The same figures every week or month, as files.
A custom cut or benchmark
Your countries, languages or fields, or your own platform against the market.
Building AI and need the experts behind this data? (you contract them yourself) or (we deliver the expert work).
Investing in AI-data companies? Hiring radar is these records role by role, every day, as files.
Method and definitions
The profiles of the people looking for AI training work on AITraining.jobs (their LinkedIn profile, CV or own words, read by the site), the credentials they state or their profile shows, and the job alerts they subscribe to. The public job boards of the AI training platforms this site tracks, read every day. Every figure is computed by one database query over the window shown, the same query for the live report and for every edition. Each figure states its sample beneath it.
- Member
- A person with a profile on AITraining.jobs by the window's end: someone looking for AI training work who gave the site their LinkedIn profile, their CV or their own description of their work.
- Field
- A member's field is the field of the best fit their profile reading names (software and IT, health and medicine, law, languages and translation, and the other groups the site's matching uses). A role is in a field when its title names one of the field's professions, by the rule the profession pages list roles with.
- Fit check
- One member against one role: the site reads the member's profile (their work, never their name or contact details) beside the role's own posting and answers strong, possible or unlikely, with what fits and what the role asks for that the background does not show. Checks are made ahead for the roles the site lists for each member and when a member asks about a role.
- Open role
- A role listed by a given day and still in its platform's feed that day.
- Open to a country
- A role is open to residents of a country if it is open worldwide, states no rule, its list names the country, or its exclusion list does not.
- Kind of work
- Each role is classified from its title and description into one of the twelve kinds of AI training work this site tracks, from AI evaluation and coding to multilingual and audio work. Every kind links to its own page.
- No cell points at a person
- No count of fewer than 3 people is published, nor any share that would reveal one: with 3 or more, no one, not even a person counted, can tell who the others are. When one hidden cell could be worked out from a row's total, a second is hidden too.
- Window
- Live: members and roles open at the end of the last complete day; fit checks and roles opened in the last 28 complete days. Edition: one calendar month.
What this data cannot tell you
- These are the people who use AITraining.jobs, not every person who does AI training work: the site's audience leans to the countries and fields it is found in.
- A profile reading can miss what a profile does not say, and a credential nobody states is Not known, never guessed.
- Platforms that recruit without a public job board are not counted. Several large ones invite workers to projects directly.
- A listing is not a hire. Some platforms keep evergreen roles open for months, and some post one listing per language or country.
- Most general rating, writing and language roles name no profession in their title, so a field's open roles are a floor and its people per role a ceiling.
How the job feeds are read and roles are classified: site methodology. How fresh, complete and consistent the data is, checked every day: data quality and coverage.
Editions
The live report covers the last 28 complete days and changes every day. Each complete month is also published as an edition, frozen two days after the month ends, so a figure you cite from an edition never moves.
- Live data (this page)Last 28 complete days, recomputed daily
Cite and reuse
Free to quote, chart and republish under CC BY 4.0 with a link to this page. Live figures change daily: cite a monthly edition for numbers that will read the same later, or keep the access date.
Citation
AITraining.jobs (2026). AI Training Talent: Supply and Demand (live data, 29 Sep 2026 to 26 Oct 2026). https://aitraining.jobs/research/ai-training-talent-supply-and-demand (accessed 27 Oct 2026)
BibTeX
@misc{aitrainingjobs_ai_training_talent_supply_and_demand_live,
author = {{AITraining.jobs}},
title = {AI Training Talent: Supply and Demand (live data, 29 Sep 2026 to 26 Oct 2026)},
year = {2026},
howpublished = {\url{https://aitraining.jobs/research/ai-training-talent-supply-and-demand}},
note = {Accessed 27 Oct 2026},
}More research
- Applicant Demand for AI Training WorkWhere do people who want AI training work apply, and which roles are short of applicants?Read the report
- How AI Training Applications WorkWhat do AI training application forms ask, and how long are they?Read the report
- What Draws Applicants to AI Training RolesWhich traits of an AI training role draw more applicants, compared within the same kind of work?Read the report
- AI Training Work Without ExperienceHow much AI training work is open to people without a degree or professional experience, and what does it pay?Read the report
- How AI Training Platforms Screen ExpertsHow do AI training platforms screen and verify the experts they hire, and how much do they ask of applicants?Read the report
- Who Looks for AI Training WorkWho is looking for AI training work: in which fields, with what credentials, from which countries, and how many already work on an AI training platform?Read the report
- Do Applicants Qualify for AI Training Roles?When the people looking for AI training work are checked against the roles that suit them best, how often do they qualify, and what do they most often lack?Read the report
- The State of AI Training PayWhat does AI training work pay, by kind of work and by platform?Read the report