Strategic Project Lead
Generalisttoday
$80-$90/hr
data annotation
514 open jobsmost pay $5–95/hr11 platforms hiringupdated today
Right now, 514 data annotation jobs are open across 11 platforms. Most pay $5–95/hr, and each one is checked at the source every 24 hours. What you earn depends on the data you label: in our study, text work paid a median $32/hr while rating search results paid $8/hr.
Updated 9 days ago ·
Country not listed? Pick ‘Any country’ and check each job’s ‘Open to’ line.
Hiring now
Generalisttoday
$80-$90/hr
Lawyesterday
$75-$95/hr
Software engineeringtoday
$10-$40/hr
$75-$95/hr
Generalisttoday
$139-$140/hr
Generalistyesterday
Not listed
Generalistyesterday
$139-$140/hr
Languages & translationyesterday
$9/hr
Generalistyesterday
$21-$35/hr
Medicine & healthcareyesterday
$100/hr
Generalistyesterday
$35-$40/hr
Medicine & healthcareyesterday
Not listed
Generalistyesterday
Not listed
Finance & accountingyesterday
$75-$95/hr
Generalistyesterday
Not listed
$75-$95/hr
Generalist2 days ago
$21-$35/hr
Medicine & healthcareyesterday
Not listed
Generalist2 days ago
$21-$35/hr
Medicine & healthcareyesterday
Not listed
Generalist3 days ago
$20-$40/hr
Generalistyesterday
Not listed
Generalist3 days ago
$60-$80/hr
Medicine & healthcareyesterday
Not listed
Generalist3 days ago
$50-$100/hr
Languages & translationyesterday
$10/hr
Generalist3 days ago
$50-$105/hr
Generalistyesterday
Not listed
Generalist3 days ago
$6-$21/hr
Generalistyesterday
Not listed
We may earn a referral fee when you’re hired through some links. How we make money
See all 514 jobs in the job board
Key takeaways
- Data annotation is the work of labelling, rating or transcribing data against written guidelines. We studied 504 data annotation listings, and 66% of them were text work: chatbot answers, AI writing, translation and language checks.
- The data sets the pay. Median advertised rates came to $32/hr for text, $24.60/hr for audio, images and video, and $8/hr for rating search, ads and maps results.
- Search, ads and maps rating is priced per country: 71 of the 83 listings were open to a single country, and the same Ads Quality Rater role ran from $2.40/hr in India up to $25/hr in Iceland.
- Gear you must own shows up in 77% of search, ads and maps listings, against just 4.5% of text listings.
- Pick AI Trainer HQ if you want every open listing's pay and countries side by side before you sit a test; DataAnnotation's general track lives on its own site and isn't posted here.
Most job boards compress this work into one line and one pay range. You can clear an unpaid test and only then discover the role is ads rating priced for your country, or that it demands a headset you don't own. So we grouped every listing by the type of data you'd handle and measured what each type pays and expects you to bring.
A data annotation job is paid work: you label, rate or transcribe data (chatbot answers, search results, audio, images or video) against written guidelines, and your judgments teach AI models. One text task asks you to pick the more accurate of two chatbot answers and say why. A search task asks whether a result matches its query; an audio task asks you to write out what a speaker said.
On this page, data annotation covers four job types: annotation, response evaluation, search rating and transcription. Platforms post the same work under different titles, so we group them together. If you're after DataAnnotation, the company, read our DataAnnotation review.
Of the 504 data annotation listings on our board on October 19, 2026, 66% would have you working with text, 17% rating search, ads or maps results, 11% with audio and 6% with images or video.
How we did it. The pool started at 2,136 live AI-training listings on 19 October 2026, once we removed 611 Mindrift paid research sessions. Within it, 504 data annotation listings spanned the four job types above, and we placed each in one data type using its title (the images and video group includes roles that record video for AI datasets).
For pay, we took the midpoint of each advertised range in US dollars per hour and report medians. Audio (57 listings) and images and video (31) each fall under our 50-listing minimum, so we combined them for the pay and equipment figures. How we collect jobs.
Share of the 504 data annotation listings by the data you'd handle, 19 October 2026.
| Data type | Listings (share) | Median pay (listings with pay) | Name equipment | Posted most by |
|---|---|---|---|---|
| Text and chatbot answers | 333 (66%) | $32/hr (170) | 4.5% | Lilt, Welo Global, micro1 |
| Search, ads and maps | 83 (17%) | $8/hr (47) | 77% | Welo Global, RWS TrainAI |
| Audio, images and video | 88 (audio 57, images and video 31) | $24.60/hr (81) | 49% | micro1, Welo Global, Mercor |
Source: AI Trainer HQ analysis of 504 data annotation listings from 19 October 2026.
Data annotation pay swings with the kind of data you label: in our study, the median for rating search, ads and maps results was a quarter of the text-work median, with audio, image and video work falling between them (the table above lists medians by type).
Median computed from each listing's advertised pay range, across 298 listings that state pay.
Text ranks highest because it includes expert written review: Mindrift's medical AI evaluation roles for physicians advertised $35 to $70 an hour on our board on 19 October 2026. Inside any single type, your field and your language shift the rate more than the platform does.
The pay-by-platform table on this page gives today's medians; for rates across every kind of AI-training work, see what AI training pays.
Search, ads and maps rating is posted one country at a time and priced for that country: 71 of the 83 listings were open to exactly one country.
Welo Global's Ads Quality Rater role shows how wide that spread runs. Across its 27 Ads Quality Rater listings, pay ran from $2.40/hr for Urdu, Telugu and Bengali raters in India to $25/hr for Icelandic raters in Iceland. India-only postings sat at a median of $4/hr across 10 India-only listings.
Search-rating listings on our job board open to applicants in India, 29 October 2026. The board displays pay rounded to whole dollars.
Before you burn an hour on a rater exam, check the country line and the rate. Rater roles have a dedicated search quality rater jobs list.
A computer covers most text work on its own, but 77% of search, ads and maps listings and 49% of audio, image and video listings name equipment you must have.
Share of listings in each data type whose requirements name equipment.
For search, ads and maps roles, the demands usually stop at reliable internet and a smartphone. Audio, image and video listings asked for a headset in 22 cases and a microphone in 20. Genuine platforms never bill you for gear, so a "recruiter" who tells you to buy equipment is following a scam pattern; see how to tell a real listing from a fake.
The right fit comes down to your native language, your gear and your country, so choose the data type before the platform. The AI Trainer HQ Data-Type Match finds it with three questions:
The AI Trainer HQ Data-Type Match.
| Data type | Best for | Trade-off | Proof |
|---|---|---|---|
| Text and chatbot answers | Strong writers, subject specialists, second-language speakers | Listed per language; many roles ask for a degree or field knowledge | 333 listings |
| Search, ads and maps | People with a phone and reliable internet who want short, repeatable tasks | Pay rate depends on your country | 83 listings, 71 for one country |
| Audio | Native speakers who own a headset and have a quiet room | Equipment is often needed | 57 listings |
| Images and video | Video, design or 3D skills, or your own camera for recording tasks | Smallest category; some roles mean filming yourself | 31 listings |
Where AI Trainer HQ lands
- Best for: checking every open data annotation listing with its $/hr and the countries that can apply, before you take an unpaid test.
- Trade-off: we employ no one and handle no applications; you apply on each platform's own site, and DataAnnotation lists nothing here.
- Proof: the 504-listing study behind this page, plus listings verified at the source every 24 hours.
Yes, for some general text, search-rating and transcription roles, but almost every platform gates paid work behind an unpaid assessment.
DataAnnotation's Starter Assessment lasts about an hour, with one attempt allowed (DataAnnotation FAQ). Outlier requires at least an associate degree (Outlier FAQ), Mercor and micro1 screen through AI interviews, and Mindrift's all-access tasks call for no education, experience or skills (Mindrift FAQ). To see who gets hired, read how to become an AI trainer or browse entry-level AI training jobs.
Pick a data type that suits you, narrow this list to your country, apply to two or three platforms, and treat every assessment like an exam.
Most listings are genuine: AI-training companies pay contractors to do real work. Scammers borrow those companies' names, and a legitimate platform never asks you for money.
The work itself follows written guidelines, and most people pick up the rules within a few hours. Clearing the entry test is the hard part: each platform scores your accuracy before paid tasks open up.
Yes, many text and audio projects hire across borders. Rating search results, ads and maps tends to be locked to a single country, so filter by yours before anything else.
No. DataAnnotation (dataannotation.tech) is a single company, and none of its openings appear on our board. For similar platforms with open listings here, check sites like DataAnnotation.
Whichever one matches your language, your equipment and your country (see the Data-Type Match). AI Trainer HQ lets you sort open listings by pay and country before you apply.
Rarely. The work is contracted and shows up in projects, which is why most annotators hold accounts on two or three platforms.
It varies by platform. Mercor pays every Wednesday (Mercor payments); micro1 and Mindrift pay twice monthly, around the 5th and 20th (micro1 payroll policy, how Mindrift pays).
Start with the list: the open data annotation jobs at the top of this page can be filtered by your country and pay band.
eligibility
pay
Median and range in US dollars per hour, from the live jobs on this page.
| Platform | Median | Range (USD/hr) | Open roles |
|---|---|---|---|
| 10x Team | $129/hr | $86–171/hr | 2 |
| Mindrift | $100/hr | $100/hr | 28 |
| Mercor | $90/hr | $30–200/hr | 32 |
| Alignerr | $85/hr | $10–120/hr | 13 |
| Meridial | $65/hr | $5–65/hr | 24 |
| micro1 | $41/hr | $2–150/hr | 105 |
| Anyone AI | $37/hr | $7–65/hr | 4 |
| Rise Data Labs | $30/hr | $5–50/hr | 10 |
| Welo Global | $11/hr | $2–105/hr | 115 |
$86–171/hr · 2 open
$129/hr median
$100/hr · 28 open
$100/hr median
$30–200/hr · 32 open
$90/hr median
$10–120/hr · 13 open
$85/hr median
$5–65/hr · 24 open
$65/hr median
$2–150/hr · 105 open
$41/hr median
$7–65/hr · 4 open
$37/hr median
$5–50/hr · 10 open
$30/hr median
$2–105/hr · 115 open
$11/hr median
platforms
Sorted by Legit Score, our 0–100 rating of how safe each platform is to work for.
Freelance expert AI trainers: surgeons, CTOs and other senior specialists.
Labelbox runs this AI-training network for domain experts and skilled generalists, with posted rates from about $10 to $200 an hour.
Developer AI-trainer roles hired by country, mostly Latin America.
AI training contributors by language, plus subject-matter experts.
Experts in law, medicine, finance and coding get matched to AI-training contracts after a résumé review, a 20-minute AI interview and assessments.
Freelance AI trainer projects by domain, run by Invisible Technologies.
An AI data lab that takes on vetted experts as remote contractors to produce training and evaluation data, with applicants screened by its AI interviewer, Zara.
Toloka runs this freelance AI-training platform: expert projects in 90+ fields alongside open "all-access" tasks anyone can start.
AI data and evaluation roles for generalists and specialists.
AI conversation specialists hired by language and dialect.
Search quality raters and AI data roles in many languages.
the work
Data annotation jobs pay you to label data that AI learns from: picking the stronger chatbot answer, tagging images, transcribing audio or marking up text to written guidelines. Across the 271 live jobs on this page that list pay, most pay $5–95/hr, and the median is $35/hr. Most of these roles are remote freelance contracts, and you set your own hours.