Skip to content

Research report ยท Live data

How AI Training Platforms Screen Experts

Of the 3,260 open roles whose full description is read (28 platforms), 12.6% state a test, interview or check, most often an assessment, test or trial task (9.2%), and 10.5% name a professional credential or licence; the 1,457 application forms read on 12 platforms require a median of 10 fields, and 92.9% require a resume.

Data window
20 Sep 2026 to 17 Oct 2026
28 days, UTC
Computed
18 Oct 2026, 16:45 UTC
Recomputed every day
Source
The public job boards of the AI training platforms this site tracks, read every day: the full description of each open role where the board publishes one, and the application form of each open role where the platform's applicant-tracking system (Greenhouse, Ashby, Lever, Workable, Teamtailor and others) is read.
Method
Version 1. Definitions
CC BY 4.0, how to cite

Every figure on this page as CSV and JSON, sent to your work email.

platforms with open roles whose descriptions or forms are read
28 of 29
of described roles state a test, interview or check
13%
required fields on a typical application form
10
of application forms require a resume
93%

Key findings

  1. 1

    Screening can be read for 28 of the 29 platforms with open roles on 17 Oct 2026: full descriptions of 3,260 roles on 28 platforms (88.9% of open roles) and application forms of 1,457 roles on 12 (39.7%); Alignerr publishes neither.

    Figure 1
  2. 2

    12.6% of the 3,260 described roles state a selection step (a test, an interview or a check), most often an assessment, test or trial task (9.2%), then an interview with a person or of an unstated kind (6.1%); 0.6% state a background check or identity verification.

    Figure 2
  3. 3

    14.6% of described roles ask for proof of expertise: 10.5% name a professional credential or licence and 4.8% ask for work samples, a portfolio or a publication record.

    Figure 2
  4. 4

    Most described roles state none of the 7 screening steps counted (75.7%); 7.6% state two or more.

    Figure 3
  5. 5

    Of the 24 platforms with enough descriptions to compare, AfterQuery states the most screening (31.9% of its 216 roles read name two or more steps) and Mindrift, Welo Data, Anyone AI, Sigma AI, Perle Labs, Appen, Babel Audio, Claru, OpenTrain AI, OneForma, Terac, Sovrano AI, Vetto and xAI (SpaceXAI) the least (none name two or more).

    Figure 4
  6. 6

    Of the 12 kinds of work with at least 10 described roles, Research Studies & Surveys roles state the most screening (18.8% name two or more steps) and Transcription & Audio and Writing roles the least (none name two or more).

    Figure 5
  7. 7

    Of the 1,457 application forms read, 92.9% require a resume, 93.7% a phone number and 46.6% a LinkedIn profile; 4.8% require an answer on work authorization, 2.5% on availability and 10.3% on pay.

    Figure 6
  8. 8

    A form has a median of 10 fields, 10 of them required; among the 10 platforms with at least 10 forms read, the median runs from 3 required fields at Terac to 20 at xAI (SpaceXAI).

    Figure 8

Figure 1

What this report can see, platform by platform

For which platforms can screening be read, and from which source?

PlatformOpen rolesFull descriptions readApplication forms read
All platforms
3,668
88.9%39.7%
Invisible
532
100.0%100.0%
Alignerr
386
0.0%0.0%
SuperAnnotate (SME)
377
100.0%0.0%
Mindrift
297
100.0%100.0%
micro1
248
100.0%0.0%
Welo Data
235
100.0%100.0%
Mercor
229
100.0%0.0%
AfterQuery
217
99.5%0.0%
Turing
193
98.4%0.0%
Handshake AI
170
100.0%2.9%
Anyone AI
128
100.0%100.0%
Sigma AI
90
100.0%100.0%
Braintrust
78
100.0%0.0%
Perle Labs
67
100.0%0.0%
TransPerfect DataForce
66
100.0%83.3%
Claru
60
70.0%0.0%
Appen
55
100.0%100.0%
Babel Audio
43
100.0%0.0%
OpenTrain AI
42
100.0%0.0%
RemoExperts
36
100.0%0.0%
OneForma
24
100.0%0.0%
Terac
22
100.0%100.0%
Sovrano AI
20
100.0%0.0%
Vetto
20
100.0%100.0%
xAI (SpaceXAI)
15
100.0%100.0%
Pasiflora AISmall sample
7
100.0%0.0%
Outlier (Scale AI)Small sample
5
100.0%0.0%
Pila8Small sample
3
100.0%0.0%
RemotasksSmall sample
3
100.0%100.0%

Sample. The 3,668 roles open on 17 Oct 2026 on the 29 platforms AITraining.jobs reads; full descriptions of 3,260 of them and application forms of 1,457.

Shares are of each platform's own open roles. A platform's board either publishes full descriptions or exposes its application form through an applicant-tracking system, and some do neither: each later figure ranks a platform only on the source it publishes.

Descriptions and forms are read on the same day as the open roles.

Source. AITraining.jobs Research, How AI Training Platforms Screen Experts, live data.

Figure 2

Screening steps stated in role descriptions

How often do platforms say they will test, interview or verify an applicant, or ask for proof of expertise?

StepRoles stating itShare of described roles
An assessment, test or trial task300
9.2%
An AI-led or recorded interview140
4.3%
An interview with a person, or of an unstated kind199
6.1%
A background check14
0.4%
Identity verification13
0.4%
A professional credential or licence named343
10.5%
Work samples, a portfolio or publications155
4.8%

Sample. 3,260 roles open on 17 Oct 2026 whose job board publishes a full description (AfterQuery, Anyone AI, Appen, Babel Audio, Braintrust, Claru, Handshake AI, Invisible, Mercor, micro1, Mindrift, OneForma, OpenTrain AI, Outlier (Scale AI), Pasiflora AI, Perle Labs, Pila8, RemoExperts, Remotasks, Sigma AI, Sovrano AI, SuperAnnotate (SME), Terac, TransPerfect DataForce, Turing, Vetto, Welo Data and xAI (SpaceXAI)): 88.9% of the 3,668 roles open on the 29 platforms read.

A step counts when the description states it, as a requirement or as something that may be asked ('a brief assessment may be added'). A platform can screen more than it writes down, so every share is a floor.

The assessment, interview and check steps are read only in the passage about applying and selection (a selection section, or a sentence that speaks of applicants, candidates, onboarding or the hiring process), so 'conduct psychological assessments' or 'transcribe interviews' in the work itself is not counted. The requirements report counts any mention, so its shares are higher.

An AI-led or recorded interview has its own line; a role that also states a further interview ('an AI interview ... and an interview with a recruiter') counts on both. A credential or licence, and work samples, a portfolio or a publication record, are proof of expertise and are read from the whole description: named as a requirement or a preference.

Invisible lists 16.3% of the described roles, so the pooled shares lean on its process; the by-platform figure separates the platforms.

Source. AITraining.jobs Research, How AI Training Platforms Screen Experts, live data.

Figure 3

Screening depth: steps stated per role

How many screening steps does a typical AI training role describe?

Steps statedRolesShare of described roles
None stated2,468
75.7%
1 step543
16.7%
2 steps132
4.0%
3 steps or more117
3.6%

Sample. 3,260 roles open on 17 Oct 2026 whose job board publishes a full description (AfterQuery, Anyone AI, Appen, Babel Audio, Braintrust, Claru, Handshake AI, Invisible, Mercor, micro1, Mindrift, OneForma, OpenTrain AI, Outlier (Scale AI), Pasiflora AI, Perle Labs, Pila8, RemoExperts, Remotasks, Sigma AI, Sovrano AI, SuperAnnotate (SME), Terac, TransPerfect DataForce, Turing, Vetto, Welo Data and xAI (SpaceXAI)): 88.9% of the 3,668 roles open on the 29 platforms read; a median of 0 of the 7 steps stated.

Screening depth: how many of the 7 steps in the steps figure a description states.

A step counts when the description states it, as a requirement or as something that may be asked ('a brief assessment may be added'). A platform can screen more than it writes down, so every share is a floor.

Source. AITraining.jobs Research, How AI Training Platforms Screen Experts, live data.

Figure 4

Screening stated, by platform

Which platforms describe the most screening, and which steps does each state?

PlatformAssessmentAI interviewInterviewBackground checkID checkMedian steps2 or more steps
All described roles9%4%6%0%0%11%5%0.08%
Invisible0%0%0%0%0%3%4%0.00%
SuperAnnotate (SME)28%27%27%0%0%15%0%0.028%
Mindrift0%0%0%0%0%11%0%0.00%
micro13%10%4%0%1%21%7%0.06%
Welo Data4%0%0%0%0%0%0%0.00%
Mercor8%1%1%0%0%18%19%0.010%
AfterQuery35%0%0%7%5%30%11%1.032%
Turing19%1%7%0%0%10%9%0.04%
Handshake AI2%0%2%0%0%7%8%0.04%
Anyone AI0%0%0%0%0%2%0%0.00%
Sigma AI0%0%0%0%0%0%0%0.00%
Braintrust28%15%28%0%0%10%0%1.017%
Perle Labs13%0%67%0%0%0%2%1.00%
TransPerfect DataForce14%0%0%0%0%5%21%0.014%
Appen0%0%0%0%0%0%0%0.00%
Babel Audio0%0%0%0%0%0%0%0.00%
Claru0%0%0%0%0%0%0%0.00%
OpenTrain AI2%0%0%0%0%0%0%0.00%
RemoExperts0%0%0%0%0%69%6%1.06%
OneForma0%0%0%0%0%0%0%0.00%
Terac0%0%0%0%0%5%5%0.00%
Sovrano AI5%0%0%0%0%30%5%0.00%
Vetto15%0%5%0%0%0%0%0.00%
xAI (SpaceXAI)0%0%0%0%0%7%0%0.00%
Pasiflora AISmall sample0%0%0%0%0%0%0%0.00%
Outlier (Scale AI)Small sample0%0%0%0%0%0%0%0.00%
Pila8Small sample0%0%0%0%0%0%0%0.00%
RemotasksSmall sample0%0%0%0%0%0%0%0.00%

Sample. 3,260 roles open on 17 Oct 2026 whose job board publishes a full description (AfterQuery, Anyone AI, Appen, Babel Audio, Braintrust, Claru, Handshake AI, Invisible, Mercor, micro1, Mindrift, OneForma, OpenTrain AI, Outlier (Scale AI), Pasiflora AI, Perle Labs, Pila8, RemoExperts, Remotasks, Sigma AI, Sovrano AI, SuperAnnotate (SME), Terac, TransPerfect DataForce, Turing, Vetto, Welo Data and xAI (SpaceXAI)): 88.9% of the 3,668 roles open on the 29 platforms read. Only these platforms publish descriptions, so only they are compared here.

Each step column is the share of the platform's described roles that state it. A platform needs 10 described roles to be read on its own.

Many descriptions on one platform share a template, so a platform's share often reflects one standard paragraph about its process rather than role-by-role choices.

A step counts when the description states it, as a requirement or as something that may be asked ('a brief assessment may be added'). A platform can screen more than it writes down, so every share is a floor.

Source. AITraining.jobs Research, How AI Training Platforms Screen Experts, live data.

Figure 5

Screening stated, by kind of work

Which kinds of AI training work are screened the hardest?

Kind of workAssessmentAI interviewInterviewBackground checkID checkMedian steps2 or more steps
Research Studies & Surveys22%0%0%0%0%3%22%0.019%
Law, Medicine & Finance11%3%4%1%1%31%8%0.011%
Coding & Software Evaluation12%11%10%0%0%3%2%0.011%
Math & STEM10%2%3%2%2%5%11%0.011%
AI Evaluation & Rating (RLHF)7%7%8%1%0%6%5%0.08%
Multilingual & Localization10%6%7%0%0%0%2%0.07%
Red-Teaming & AI Safety3%3%3%0%0%3%8%0.06%
Data Annotation & Labeling9%1%2%0%0%2%6%0.04%
Robotics & Physical-AI Data2%4%0%0%4%0%0%0.04%
Agentic Tasks & RL Environments5%2%5%0%0%0%2%0.02%
Transcription, Audio & Speech4%0%11%0%0%0%1%0.00%
Writing & Content0%0%0%0%0%4%4%0.00%

Sample. 3,260 roles open on 17 Oct 2026 whose job board publishes a full description (AfterQuery, Anyone AI, Appen, Babel Audio, Braintrust, Claru, Handshake AI, Invisible, Mercor, micro1, Mindrift, OneForma, OpenTrain AI, Outlier (Scale AI), Pasiflora AI, Perle Labs, Pila8, RemoExperts, Remotasks, Sigma AI, Sovrano AI, SuperAnnotate (SME), Terac, TransPerfect DataForce, Turing, Vetto, Welo Data and xAI (SpaceXAI)): 88.9% of the 3,668 roles open on the 29 platforms read.

A kind's row pools the platforms that publish descriptions. 'From one platform' is the share of its roles listed by the platform with the most of them: near 100%, the row reads that platform's process.

Rows with fewer than 10 roles read are marked thin.

Source. AITraining.jobs Research, How AI Training Platforms Screen Experts, live data.

Figure 6

What application forms require

Which documents and answers must an applicant give before the form will submit?

ItemForms askingForms requiring it
Resume or CV98.4%
92.9%
Phone number99.3%
93.7%
Location or country95.5%
85.0%
LinkedIn profile65.2%
46.6%
Portfolio, website or GitHub25.8%
8.0%
Work authorization or visa sponsorship4.8%
4.8%
Availability, hours or start date3.1%
2.5%
Expected pay or rate10.3%
10.3%

Sample. 1,457 application forms of roles open on 17 Oct 2026, read from the applicant-tracking systems of Anyone AI, Appen, Handshake AI, Invisible, Mindrift, Remotasks, Sigma AI, Terac, TransPerfect DataForce, Vetto, Welo Data and xAI (SpaceXAI): 39.7% of the 3,668 roles open on the 29 platforms read.

Required: the form will not submit without it. Name and email, asked by every form, are not listed.

Work authorization, availability and pay are read from the label of each question the platform adds (the applications report's question themes). The pay theme covers both questions asking an applicant's expected rate and questions asking them to accept the stated rate.

Invisible lists 36.5% of the forms, so the pooled shares lean on its process; the by-platform figure separates the platforms.

Source. AITraining.jobs Research, How AI Training Platforms Screen Experts, live data.

Figure 7

Form burden: required fields per form

How much must an applicant fill in before a form will submit?

Required fieldsFormsShare of forms
4 or fewer123
8.4%
5 to 9474
32.5%
10 to 14771
52.9%
15 to 1976
5.2%
20 or more13
0.9%

Sample. 1,457 application forms of roles open on 17 Oct 2026, read from the applicant-tracking systems of Anyone AI, Appen, Handshake AI, Invisible, Mindrift, Remotasks, Sigma AI, Terac, TransPerfect DataForce, Vetto, Welo Data and xAI (SpaceXAI): 39.7% of the 3,668 roles open on the 29 platforms read; a median of 10 required fields of 10 in all.

Form burden: the fields a form will not submit without, from name and email to the last required screening question. Some platforms apply in steps (a short form, then an assessment); only the first form is read.

Source. AITraining.jobs Research, How AI Training Platforms Screen Experts, live data.

Figure 8

Form burden, by platform

Which platforms ask the most of applicants on the form itself?

PlatformMedian required fieldsMedian fieldsResume requiredPhone required
All forms read
10.0
10.093%94%
xAI (SpaceXAI)
20.0
25.0100%100%
Handshake AISmall sample
16.0
18.0100%100%
TransPerfect DataForce
15.0
18.093%100%
Vetto
13.0
22.05%95%
RemotasksSmall sample
11.0
12.0100%100%
Sigma AI
11.0
17.099%70%
Anyone AI
10.0
10.0100%93%
Invisible
10.0
10.099%100%
Welo Data
8.0
15.0100%100%
Appen
7.0
13.00%0%
Mindrift
6.0
7.0100%100%
Terac
3.0
3.00%100%

Sample. 1,457 application forms of roles open on 17 Oct 2026, read from the applicant-tracking systems of Anyone AI, Appen, Handshake AI, Invisible, Mindrift, Remotasks, Sigma AI, Terac, TransPerfect DataForce, Vetto, Welo Data and xAI (SpaceXAI): 39.7% of the 3,668 roles open on the 29 platforms read. Only these platforms' forms are read, so only they are compared here.

Work authorization, availability and pay: the share of forms with a required question of that theme.

A platform needs 10 forms read to be read on its own. Forms of one platform usually share one template, so its medians describe that template.

Source. AITraining.jobs Research, How AI Training Platforms Screen Experts, live data.

Figure 9

Screening questions on the form, by platform

What does each platform's form ask applicants about, beyond the resume?

PlatformWork authorization or visa sponsorshipAvailability, hours or start dateExpected pay or rateDegree or field of studyExperience or years in a fieldLanguages spoken or fluencyLicence or certificationPrior AI training or annotation workAge or being over 18
All forms read5%3%10%11%10%25%0%2%0%
Anyone AI0%0%81%0%0%94%0%0%0%
Appen0%0%0%100%0%100%0%0%0%
Handshake AISmall sample100%20%0%0%0%0%0%0%0%
Invisible0%2%1%0%0%4%0%0%0%
Mindrift0%0%0%0%0%0%0%0%0%
RemotasksSmall sample67%0%0%100%33%33%0%0%0%
Sigma AI1%0%0%70%26%70%0%3%0%
Terac0%0%0%0%0%0%0%0%0%
TransPerfect DataForce29%16%60%9%49%56%0%2%7%
Vetto0%5%0%5%40%70%0%0%0%
Welo Data13%11%3%16%38%21%0%12%0%
xAI (SpaceXAI)100%0%0%7%0%87%0%0%0%

Sample. 1,457 application forms of roles open on 17 Oct 2026, read from the applicant-tracking systems of Anyone AI, Appen, Handshake AI, Invisible, Mindrift, Remotasks, Sigma AI, Terac, TransPerfect DataForce, Vetto, Welo Data and xAI (SpaceXAI): 39.7% of the 3,668 roles open on the 29 platforms read; themes read from the label of each question the platform adds to its form.

Each cell is the share of the platform's forms asking at least one question of the theme. A question can match more than one theme.

Questions on how the applicant heard of the role and voluntary equal-opportunity questions are left out: they do not screen.

Source. AITraining.jobs Research, How AI Training Platforms Screen Experts, live data.

Figure 10

Form burden, by kind of work

Do forms for some kinds of AI training work ask more of applicants?

Kind of workMedian required fieldsMedian fieldsResume requiredPhone required
Red-Teaming & AI SafetySmall sample
14.0
21.033%100%
Data Annotation & Labeling
12.0
18.093%96%
Coding & Software Evaluation
10.0
10.095%99%
Law, Medicine & Finance
10.0
10.091%97%
Multilingual & Localization
10.0
10.098%95%
AI Evaluation & Rating (RLHF)
10.0
14.088%92%
Math & STEM
10.0
10.098%99%
Writing & ContentSmall sample
10.0
10.080%100%
Transcription, Audio & Speech
7.0
11.076%76%
Agentic Tasks & RL Environments
6.0
7.073%100%
Research Studies & Surveys
6.0
7.079%100%
Robotics & Physical-AI DataSmall sample
6.0
12.575%75%

Sample. 1,457 application forms of roles open on 17 Oct 2026, read from the applicant-tracking systems of Anyone AI, Appen, Handshake AI, Invisible, Mindrift, Remotasks, Sigma AI, Terac, TransPerfect DataForce, Vetto, Welo Data and xAI (SpaceXAI): 39.7% of the 3,668 roles open on the 29 platforms read.

A kind's row pools the platforms whose forms are read. 'From one platform' is the share of its forms from the platform with the most of them: near 100%, the row reads that platform's form.

Rows with fewer than 10 forms read are marked thin.

Source. AITraining.jobs Research, How AI Training Platforms Screen Experts, 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:

  • 9,572 roles tracked, each with its history

    Every role on 29 job boards, read daily since 13 Jul 2026: its pay and place, and the first and last day it was open.

  • 6,200 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.

Method and definitions

The public job boards of the AI training platforms this site tracks, read every day: the full description of each open role where the board publishes one, and the application form of each open role where the platform's applicant-tracking system (Greenhouse, Ashby, Lever, Workable, Teamtailor and others) is read. 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.

Screening step
One of seven checks a description can state: an assessment, test or trial task; an AI-led or recorded interview; an interview with a person or of an unstated kind; a background check; identity verification (the five selection steps); a professional credential or licence; work samples, a portfolio or a publication record (the two kinds of proof of expertise). A step counts when the description states it, as a requirement or as something that may be asked.
Selection passage
The part of a description about applying and being selected: a section headed as the process ('Process', 'Application steps', 'What to expect'), a sentence about the application, onboarding, screening or selection (and the sentence after it), or a sentence about the applicants outside the sections that describe the work or the qualifications. The five selection steps are read only there, with the requirements report's patterns and a few documented additions, so 'conduct psychological assessments' or 'transcribe interviews' in the work is not counted as a check.
Proof of expertise
A professional credential or licence the description names (medical, legal, nursing, financial, board certification or any licence), or work the applicant must show: samples, a portfolio or reel, a verifiable GitHub profile, or a publication record. Read from the whole description, named as a requirement or a preference.
Screening depth
How many of the seven screening steps a description states. Shown as a median and as the share of roles stating two or more.
Form burden
The number of fields an application form will not submit without, from name and email to the last required screening question.
Required
The form will not submit without the field.
Question theme
What a question the platform adds to its form is about (work authorization, availability, pay, degree, experience, languages, a licence, prior AI training work, age), read from its label with the applications report's documented patterns. The pay theme covers both an applicant's expected rate and accepting the stated rate.
Coverage
A platform's board either publishes full descriptions or exposes its application form, and some do neither. Each figure states which platforms and what share of all open roles it reads, and a platform is only ever compared on the source it publishes.
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 roles open at the end of the last 28 complete days, in UTC. Edition: the roles open at the end of one calendar month.

What this data cannot tell you

  • Stated, not observed: a platform can screen more than its descriptions say, so every share is a floor. Pass rates are not published, so this report cannot say how selective any step is.
  • Descriptions and forms come from different platforms, and some platforms publish neither: the coverage figure names them. The platforms that publish full descriptions are expert networks, so the description figures lean towards expert work.
  • Most descriptions and forms on one platform follow one template, so a platform's figure often reflects one standard paragraph or one form, and pooled shares lean on the platform with the most open roles.
  • Only the first application form is read; assessments and interviews that follow on a platform's own site are visible only when the description states them.
  • A step phrased in an unusual way can be missed, and a step stated inside a section that describes the work, without a word about applying or selection, is not counted.
  • 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.

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

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). How AI Training Platforms Screen Experts (live data, 20 Sep 2026 to 17 Oct 2026). https://www.aitraining.team/research/how-ai-training-platforms-screen-experts (accessed 18 Oct 2026)

BibTeX

@misc{aitrainingjobs_how_ai_training_platforms_screen_experts_live,
  author = {{AITraining.jobs}},
  title = {How AI Training Platforms Screen Experts (live data, 20 Sep 2026 to 17 Oct 2026)},
  year = {2026},
  howpublished = {\url{https://www.aitraining.team/research/how-ai-training-platforms-screen-experts}},
  note = {Accessed 18 Oct 2026},
}
Data: every figure as CSV (one line per value) and JSON, sent to your work email.Pay and open-roles data

More research

All 78 research reports