AfterQuery is a Y Combinator-backed company that turns professional expertise into training data for frontier AI labs, drawing on developers, attorneys and other specialists working as contractors. Applicants take one unpaid assessment per expert pool, verify their identity, and are paid weekly through Stripe to a bank account in a Stripe-supported country.
Accepts worldwide
Alignerr is Labelbox's expert network: subject-matter specialists are matched to projects from AI labs, creating training data and rating model output. Work is fully remote, and Alignerr says its experts work from almost anywhere in the world. Unusually for this market, every listing publishes its own rate before you apply.
Accepts worldwide
Appen is an ASX-listed AI data company founded in Sydney in 1996. Its contributors now work through CrowdGen by Appen, which replaced the older Appen Connect portal, on language, speech, evaluation and expert projects matched to their location and skills.
Accepts worldwide
Braintrust is a talent network for contract and full-time work that also runs AI training gigs: rating model responses, labelling data, expert evaluation, prompt writing, red teaming and code review. It charges talent no fees, and every applicant goes through an AI skills interview and ID verification before being matched to work.
Accepts worldwide
Clickworker is a German crowdworking platform, now a subsidiary of LXT, where freelancers do paid microtasks: surveys, text writing, web research, and photo, video and audio recording used as AI training data, plus Microsoft UHRS tasks for those who pass its assessment. Pay is per task, not per hour.
Accepts worldwide
DataAnnotation pays contractors to evaluate chatbot answers, write prompts and review code for AI training. It is run by Surge Labs, Inc. (Surge AI), screens with a single unpaid Starter Assessment, and publishes starting rates by track but no list of eligible countries.
Accepts worldwide
DataForce is the AI data arm of TransPerfect, the language services company. Its contributor community collects speech, text, photo and video data, annotates it, and takes part in user studies, remotely or on site. Work is project by project, often tied to a language or city, and remote projects pay only after a quality check.
Accepts worldwide
Ethos is an expert network that matches credentialed professionals to paid advisory calls, consulting projects, and AI training work for research firms, consultancies and AI labs. Joining is free and you set your own availability, but Ethos publishes no rate card and no payout terms.
Accepts worldwide
Handshake AI is the AI-training arm of the student careers network Handshake. It recruits students, graduates and credentialed specialists to train and evaluate large language models for AI labs. Work is remote and part-time, and Handshake's team matches you to a project rather than letting you pick one from a board. No prior AI experience is required.
United States
iMerit is a San Jose-based AI data company with more than 5,000 employees and, since July 2025, iMerit Scholars: an invitation-led network of contract experts who tune and evaluate generative AI models in fields from medicine and law to linguistics and music. Scholars work to a Statement of Work and submit weekly timesheets.
Accepts worldwide
Innodata is a Nasdaq-listed data engineering company that recruits part-time, flexible-hours AI evaluators, annotators and generative AI associates for its Subject Matter Expert on Demand programme. Postings are almost all US-based and state their hourly rate.
Accepts worldwide
Invisible Technologies trains and evaluates AI models for frontier labs and enterprises, and hires independent contractors for that work through Meridial, its expert marketplace. Projects cover coding, languages, voice and audio, and technical auditing. Each listing shows a fixed rate and a Statement of Work; contract terms name weekly payment via Stripe or Wise.
Accepts worldwide
Lionbridge is a translation and localisation company that recruits freelancers for AI data work (annotation, prompt writing, response evaluation) and for game testing through Lionbridge Games. Its original Lionbridge AI crowd was sold to TELUS in 2021, and the TrainAI community belongs to RWS, not Lionbridge.
Accepts worldwide
LXT is a Toronto-based AI training data company that collects and annotates speech, text, image and video data. Its remote crowd work runs through clickworker, which LXT agreed to acquire in 2024, so contributors sign up, work and get paid on clickworker's platform.
Accepts worldwide
Mercor matches domain experts to AI labs that need human judgement for training and evaluating models. Work is contract-based and remote, and projects are assigned to you by Mercor rather than picked from a board: you apply once and are put forward for whatever fits your background.
Accepts worldwide
micro1 is a data lab that pays domain experts to train and evaluate frontier AI models: engineering, law, medicine, finance, linguistics and more. Work is fully remote, entry is through an AI interview with its own interviewer Zara, and micro1 publishes the twenty countries it cannot hire from.
Accepts worldwide
Mindrift is Toloka's AI training platform, where contributors write prompts, evaluate model answers and improve responses for AI companies. Open all-access tasks need no application; specialised "AI tutor" projects need a CV, assessments and usually a relevant degree. Former Toloka contributor accounts moved here.
Accepts worldwide
OneForma is the crowd work platform operated by Centific (formerly Pactera EDGE), offering annotation, data collection, judging, transcription and translation projects in 300+ languages. You see a project's rate before accepting it, but contractor reviews of support and account suspensions are poor.
Accepts worldwide
Outlier is Scale AI's platform for freelance AI trainers: you write hard prompts, build grading rubrics, and rank model answers in your own field. Fully remote with no minimum hours, and you see a project's rate before you start it, but Outlier requires at least an associate degree, and publishes no list of the countries it operates in.
Accepts worldwide
Prolific is an Oxford-founded research platform where verified participants are paid to take part in academic and commercial studies, including AI evaluation and training tasks. The AI work goes to invited AI taskers and verified Domain Experts; everyone else starts on ordinary studies, after a waitlist and ID check.
Australia · Austria · Bahrain · Belgium · Brazil · Bulgaria · Canada · Chile · Croatia · Cyprus · Czechia · Denmark · Egypt · Estonia · Finland · France · Germany · Greece · Hong Kong SAR China · Hungary · Iceland · India · Indonesia · Ireland · Israel · Italy · Japan · South Korea · Latvia · Lithuania · Luxembourg · Malta · Mexico · Netherlands · New Zealand · Norway · Poland · Portugal · Qatar · Romania · Saudi Arabia · Singapore · Slovakia · Slovenia · South Africa · Spain · Sweden · Switzerland · Taiwan · United Arab Emirates · United Kingdom · United States
Remotasks is Scale AI's task platform for image labelling, audio transcription and data annotation. It still operates in 2026, but Scale runs its expert language-model work through Outlier instead, and Remotasks has withdrawn from some countries.
Accepts worldwide
Rise Data Labs supplies expert-made training and evaluation data to AI labs and recruits subject-matter experts, coders and professionals for contract projects. Applicants go through an AI interview and a domain skills assessment, and each role lists its own hourly rate.
Accepts worldwide
Scale AI is the San Francisco data company behind Outlier and Remotasks. Most Scale AI contract work is done through those two platforms rather than under the Scale name; Scale itself hires a small number of expert fellows directly. Meta owns a 49 percent non-voting stake.
Accepts worldwide
SuperAnnotate is an AI data platform whose expert network hires through SME Careers: doctors, lawyers, engineers, linguists and other specialists who annotate, evaluate and QA training data for AI companies. Applicants take a recorded AI interview, contract through Deel, and start each project with a paid bootcamp. Hourly work is tracked with monitoring software.
Accepts worldwide
Surge AI is a San Francisco data labelling and RLHF company, founded in 2020, that sells human training data to AI labs. Its annotator and AI trainer work runs through DataAnnotation, a platform Surge itself describes as its own; specialists in finance, healthcare, law and coding can apply through its consultant network.
Accepts worldwide
The TELUS Digital AI Community, formerly TELUS International AI, is TELUS Digital's network of freelance search raters, annotators, linguists and domain experts. Work is posted as individual roles, often tied to a country and language, with qualification exams before you start.
Accepts worldwide
Toloka is an Amsterdam-based AI training data company, founded in 2014, that sells human-generated data to AI labs. Its crowd work no longer runs under the Toloka name: tasks, accounts and balances have moved to Mindrift, Toloka's contributor platform.
Accepts worldwide
Turing is a San Francisco company that recruits software developers and domain experts to train and evaluate large language models for AI labs. Work is contract-based: you build a profile, pass proctored tests and interviews, and are matched to a project with set hours and a fixed length.
Accepts worldwide
Uber AI Solutions is Uber's AI data business, and it pays gig workers for data labelling, AI response evaluation, translation, product testing and data collection through a web platform and, in pilots, the driver app. No minimum education or experience is required.
Accepts worldwide
Welocalize is a language services company whose AI training data arm, Welo Data, hires freelance raters, annotators and language specialists for search, ads, safety and LLM evaluation work. Most roles are task-based, tied to one language and country, and screened by assessment and exam.
Accepts worldwide
Every listing shows its pay and who it is open to.