Key findings

1
17.3% of the 4,000 AI training roles with a full description mention a PhD, and 25.2% mention a PhD or a master's degree.
Figure 1
2
16.6% mention a professional credential: 124 roles a medical degree and 105 a law degree or bar admission.
Figure 2
3
Where a role states years of experience (26.0% of roles), the median ask is 3 years; 27.8% ask for 5 years or more.
Figure 3
4
Roles that state weekly hours (24.5%) ask for a median of 15 hours a week; 22.6% ask for 40 or more.
Figure 4
5
20.0% of roles mention an assessment, test or trial task, and 5.0% an AI-led or recorded video interview.
Figure 5
6
The tools named most often are Python (462 roles), Excel or spreadsheets (258 roles) and CAD tools (123 roles).
Figure 7

Figure 1

Degrees mentioned, by kind of work

How many AI training roles ask for a PhD, a master's or a degree?

Kind of work	PhD	Master's	Bachelor's	No degree named	Roles read
All roles read	17.3%	7.9%	13.5%	61.1%	4,000
Math & STEM	63.4%	12.9%	4.3%	19.4%	232
Agentic Tasks & RL Environments	37.5%	1.7%	5.8%	55.0%	120
Research Studies & Surveys	25.0%	0.0%	12.5%	62.5%	16
Law, Medicine & Finance	23.7%	9.1%	12.9%	53.5%	1,445
Coding & Software Evaluation	15.8%	8.9%	18.1%	56.8%	259
Red-Teaming & AI Safety	13.2%	2.6%	23.7%	60.5%	38
AI Evaluation & Rating (RLHF)	13.1%	10.1%	27.8%	49.0%	198
Multilingual & Localization	7.9%	11.2%	18.5%	62.5%	840
Data Annotation & Labeling	6.9%	2.8%	16.6%	73.7%	217
Writing & Content	4.0%	8.0%	28.0%	60.0%	25
Transcription, Audio & Speech	0.0%	0.8%	4.2%	95.1%	526
Robotics & Physical-AI Data	0.0%	0.0%	3.6%	96.4%	84

Sample. 4,000 roles open on 31 Oct 2026 whose job board publishes a full description (AfterQuery, Aligned Labs, Anthropic, Anyone AI, Appen, Babel Audio, Braintrust, Claru, Cohere, DATAmundi, DataAnnotation, Edison Scientific, Elsevier Health, EvolutionIQ, Fleet AI, Handshake AI, Headway, Innodata, Invisible, Inworld AI, JazzX AI, Kastle, Lifted (Upwork), Limbic, Mercor, Mindrift, Mistral AI, OneForma, OpenTrain AI, Outlier (Scale AI), Pasiflora AI, Perle Labs, Pila8, PointClickCare, Prolific, RWS TrainAI Community, RealTrainAI, RemoExperts, Remotasks, Rise Data Labs, Sama, Scale AI Human Frontier Collective, Shaip (Collaborate by Shaip), Sigma AI, Snorkel AI Expert Community, Sovrano AI, SuperAnnotate (SME), Surge AI, TELUS Digital AI, Terac, TransPerfect DataForce, Trellis, Turing, Vals AI, Vetto, Volga Partners, Weekday, Welo Data, iMerit, micro1 and xAI (SpaceXAI)), 91.3% of all open roles.

Each role counts once, at the highest degree its description mentions. A mention can be a requirement or a preference ('a PhD is a plus'): the figure counts both.

Rows with fewer than 6 described roles are marked as a small sample and not ranked: below that the average of the rows weighs more than a row's own described roles (the weight fitted on the rows' own counts).

Source. AITraining.jobs Research, What AI Training Roles Require, October 2026 edition.

Figure 2

Professional credentials mentioned

How many roles ask for a medical, legal, financial or nursing credential?

Credential	Roles mentioning it	Share of roles read	Main kind of work
A professional licence (any)	430	10.8%	Law, Medicine & Finance
Medical degree (MD or DO)	124	3.1%	Law, Medicine & Finance
Law degree or bar admission	105	2.6%	Law, Medicine & Finance
Board certification	94	2.4%	Law, Medicine & Finance
CPA or CFA	102	2.6%	Law, Medicine & Finance
Nursing licence (RN, NP)	39	1.0%	Law, Medicine & Finance

A role can mention several credentials and counts once for each.

Source. AITraining.jobs Research, What AI Training Roles Require, October 2026 edition.

Figure 3

Years of experience asked for

How much experience do AI training roles ask for?

Experience stated	Roles	Share of roles stating years
1 to 2 years	305	29.4%
3 to 4 years	444	42.8%
5 to 9 years	197	19.0%
10 years or more	92	8.9%

Sample. 1,038 of the 4,000 roles read state a number of years of experience (26.0%).

The first figure stated before 'years of experience' counts ('3+ years of clinical experience' is 3).

Source. AITraining.jobs Research, What AI Training Roles Require, October 2026 edition.

Figure 4

Weekly hours stated

How many hours a week do AI training roles expect?

Hours stated	Roles	Share of roles stating hours
Under 10 hours a week	144	14.7%
10 to 19 hours	385	39.2%
20 to 29 hours	220	22.4%
30 to 39 hours	10	1.0%
40 hours or more	222	22.6%

Sample. 981 of the 4,000 roles read state weekly hours (24.5%).

The lower figure of a stated range counts ('10 to 20 hours a week' is 10): it is the commitment a role asks for at least.

Source. AITraining.jobs Research, What AI Training Roles Require, October 2026 edition.

Figure 5

Selection steps mentioned

What does it take to get accepted?

Step	Roles mentioning it	Share of roles read
An assessment, test or trial task	798	20.0%
An interview (with a person or unspecified)	196	4.9%
An AI-led or recorded video interview	199	5.0%
A background check	20	0.5%
Identity verification	35	0.9%

An AI-led or recorded video interview is counted on its own line, not again as an interview.

Source. AITraining.jobs Research, What AI Training Roles Require, October 2026 edition.

Figure 6

Terms of work mentioned

Are AI training roles remote, part-time and freelance?

Term	Roles mentioning it	Share of roles read
Remote work	3,045	76.1%
Contractor or freelance terms	2,096	52.4%
Part-time	565	14.1%
Full-time	347	8.7%
A flexible or self-set schedule	591	14.8%
Employee benefits or W-2	623	15.6%

Source. AITraining.jobs Research, What AI Training Roles Require, October 2026 edition.

Figure 7

Tools and technical skills named

Which tools do AI training roles ask applicants to know?

Tool or skill	Roles naming it	Share of roles read	Of them, coding roles
Python	462	11.6%	24.2%
Excel or spreadsheets	258	6.5%	3.5%
CAD tools	123	3.1%	3.3%
Git or GitHub	123	3.1%	31.7%
JavaScript or TypeScript	99	2.5%	89.9%
Linux or the command line	87	2.2%	20.7%
Java	80	2.0%	88.8%
SQL	78	2.0%	35.9%
C or C++	64	1.6%	62.5%
Design tools (Figma, Adobe)	58	1.5%	1.7%
Machine learning frameworks	36	0.9%	38.9%
React	27	0.7%	85.2%
Cloud (AWS, GCP, Azure)	25	0.6%	52.0%
LaTeX	24	0.6%	0.0%
Statistics software (R, Stata, SPSS)	9	0.2%	11.1%

Source. AITraining.jobs Research, What AI Training Roles Require, October 2026 edition.

Figure 8

Requirements by platform

How do the platforms that publish full descriptions differ in what they ask?

Platform	PhD or master's mentioned	Credential mentioned	Median years asked	Median weekly hours	Assessment mentioned	Roles read
Invisible	50.5%	3.0%	2	n/a	3.8%	533
SuperAnnotate (SME)	35.6%	15.3%	3	15	11.9%	379
micro1	10.3%	22.2%	5	10	26.2%	302
Mercor	19.0%	27.0%	3	20	12.7%	237
AfterQuery	28.8%	28.4%	2	10	38.4%	229
Handshake AI	29.3%	13.0%	2	5	4.9%	184
Welo Data	0.5%	0.5%	1	20	8.2%	184
Turing	32.6%	11.4%	3	40	43.4%	175
RWS TrainAI Community	3.5%	93.0%	5	5	6.3%	143
Anyone AI	72.2%	2.3%	3	20	55.6%	133
Prolific	13.7%	11.3%	3	n/a	100.0%	124
DataAnnotation	3.6%	8.0%	2	n/a	3.6%	112
Rise Data Labs	15.7%	8.3%	5	40	5.6%	108
Weekday	30.6%	26.9%	2	20	25.0%	108
Mindrift	45.6%	44.4%	n/a	10	0.0%	90
Sigma AI	0.0%	0.0%	30	n/a	0.0%	81
Braintrust	20.8%	14.3%	3	20	27.3%	77
Perle Labs	1.5%	0.0%	n/a	n/a	16.4%	67
TELUS Digital AI	6.0%	1.5%	2	n/a	14.9%	67
Volga Partners	54.2%	54.2%	4	n/a	61.0%	59
Babel Audio	0.0%	0.0%	n/a	n/a	0.0%	47
Outlier (Scale AI)	0.0%	0.0%	n/a	40	97.8%	46
Lifted (Upwork)	4.4%	2.2%	4	15	31.1%	45
TransPerfect DataForce	2.3%	2.3%	n/a	40	2.3%	44
Appen	0.0%	0.0%	n/a	10	0.0%	42
Claru	2.4%	0.0%	n/a	n/a	4.8%	42
OpenTrain AI	12.5%	2.5%	5	20	7.5%	40
Terac	10.8%	8.1%	n/a	n/a	10.8%	37
Sovrano AI	11.1%	41.7%	3	10	55.6%	36
RemoExperts	28.1%	81.3%	3	n/a	6.3%	32
Innodata	71.0%	0.0%	4	n/a	0.0%	31
Vetto	10.0%	0.0%	3	n/a	20.0%	20
OneForma	0.0%	0.0%	n/a	n/a	5.9%	17
xAI (SpaceXAI)	17.6%	5.9%	n/a	n/a	0.0%	17
Aligned Labs	86.7%	13.3%	3	n/a	0.0%	15
Surge AI	7.7%	0.0%	2	10	7.7%	13
Scale AI Human Frontier Collective	50.0%	40.0%	n/a	10	10.0%	10
iMerit	0.0%	0.0%	n/a	n/a	11.1%	9
Shaip (Collaborate by Shaip)	0.0%	11.1%	n/a	n/a	11.1%	9
DATAmundi	0.0%	0.0%	n/a	n/a	0.0%	8
Pasiflora AI	42.9%	0.0%	n/a	n/a	14.3%	7
Elsevier Health	16.7%	83.3%	2	5	66.7%	6
Cohere	25.0%	0.0%	3	16	75.0%	4
Pila8	0.0%	0.0%	n/a	n/a	0.0%	4
RealTrainAI	25.0%	25.0%	n/a	n/a	0.0%	4
Mistral AI	33.3%	0.0%	n/a	n/a	0.0%	3
Remotasks	0.0%	0.0%	n/a	n/a	33.3%	3
Sama	0.0%	33.3%	n/a	n/a	0.0%	3
Fleet AI	0.0%	50.0%	n/a	n/a	50.0%	2
Anthropic	100.0%	0.0%	n/a	n/a	0.0%	1
Edison Scientific	100.0%	0.0%	n/a	n/a	0.0%	1
EvolutionIQ	0.0%	0.0%	n/a	n/a	0.0%	1
Headway	0.0%	100.0%	n/a	n/a	100.0%	1
Inworld AI	0.0%	0.0%	n/a	n/a	0.0%	1
JazzX AI	0.0%	0.0%	n/a	n/a	100.0%	1
Kastle	0.0%	0.0%	n/a	n/a	0.0%	1
Limbic	0.0%	100.0%	n/a	n/a	0.0%	1
PointClickCare	0.0%	100.0%	n/a	n/a	100.0%	1
Snorkel AI Expert Community	100.0%	0.0%	n/a	n/a	0.0%	1
Trellis	0.0%	100.0%	n/a	n/a	0.0%	1
Vals AI	100.0%	100.0%	n/a	n/a	100.0%	1

Medians need at least 5 roles stating the figure.

Source. AITraining.jobs Research, What AI Training Roles Require, October 2026 edition.

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:

11,482 roles tracked, each with its history
Every role on 62 job boards, read since 13 Jul 2026 (every hour since 18 Oct 2026): its pay and place, and the first and last day it was open.
58,632 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 public job boards of the AI training platforms this site tracks, read every hour. Full descriptions are read for the roles whose feed publishes them. 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.

Mention: A description mentions a degree, credential, step or tool when it contains one of the documented patterns for it (whole words, case-insensitive). A mention may be a requirement or a preference.
Highest degree: Each role counts once, at the highest degree its description mentions: PhD, then master's, then bachelor's, then student.
Years and hours: The first figure before 'years of experience', and the lower figure of a stated weekly range ('10 to 20 hours a week' is 10).
Coverage: Only roles whose job board publishes a full description are read; each figure states how many that is.
Open role: A role listed by a given day and still in its platform's feed that day.
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.
Window: Live: the last 28 complete days, in UTC. Edition: one calendar month, in UTC.

What this data cannot tell you

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.
The platforms that publish full descriptions are expert networks, so these roles lean towards expert work.
A pattern can miss a requirement phrased in an unusual way; nothing is inferred beyond what the text says.

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 dataLast 28 complete days, recomputed daily
October 2026 edition (this page)Freezes on first reading

Cite and reuse

Free to quote, chart and republish under CC BY 4.0 with a link to this page. This edition is frozen, so its figures will read the same later.

Citation

AITraining.jobs (2026). What AI Training Roles Require, October 2026 edition. https://aitraining.jobs/research/what-ai-training-roles-require/2026-10

BibTeX

@misc{aitrainingjobs_what_ai_training_roles_require_2026_10,
  author = {{AITraining.jobs}},
  title = {What AI Training Roles Require, October 2026 edition},
  year = {2026},
  howpublished = {\url{https://aitraining.jobs/research/what-ai-training-roles-require/2026-10}},
}

Data: every figure as CSV (one line per value) and JSON, sent to your work email.Pay and open-roles data

More research

All 81 research reports