Paid Expert

Expert platforms, reviewed

What each platform does, who it's open to, the kind of work on offer, and the honest pros and cons.

Introduction

Dozens of companies now pay experts to train and evaluate AI models, often at US$50–$250+/hr for specialist work. They vary a lot in how you get in, where they hire, what they pay and how reliably they pay it. This guide walks through them one by one so you can pick where to spend your time.

The first four platforms (Mercor, micro1, Ethos and DataAnnotation) have dedicated apply links that go through our referral, which supports this site at no cost to you. The rest link straight to the platform so you can explore and apply directly.

The pros and cons below summarise what contributors have said on Reddit over roughly the last three months. They reflect community sentiment, not guarantees — your own experience may differ, so always verify current terms on each platform.

Mercor

Apply to Mercor →

About

Mercor is an AI talent marketplace that matches contractors to projects for major AI labs — reinforcement learning from human feedback (RLHF), model evaluation, coding, STEM and domain-expert work. You complete an AI interview once, and Mercor often matches you to roles automatically from your profile. Advertised average pay is around $100/hr, with real project rates commonly reported between $55 and $120+/hr.

Countries served

Global, with eligibility defaulting to the US, Canada, UK and EU where a project doesn't specify otherwise.

Areas of expertise

This is just a sample, please explore the platform as there could be something perfect for you if not listed here.

Pros & cons

Pros

  • Payments are generally reliable and on time, usually weekly, and onboarding or qualification time is often paid.
  • Competitive rates — contributors report $70–$120/hr on good projects, and some have earned five figures in a few weeks.
  • Smooth onboarding and a flexible schedule that works well as a side gig.
  • Automatic matching can land you an offer without even applying.

Cons

  • Projects can end abruptly and are often short, so it's unreliable as a sole income — most say don't quit your day job.
  • Some roles advertised as hourly turn out to be pay-per-task, where you're only paid for accepted tasks, which several felt wasn't clear upfront.

Back to top ↑

micro1

Apply to micro1 →

About

micro1 is an AI hiring platform that screens experts with an AI recruiter ("Zara") and places them on expert pipelines for coding, reasoning, STEM and domain evaluation. It hires worldwide and pays per hour or per task depending on the role.

Countries served

Global — micro1 hires from 150+ countries.

Areas of expertise

This is just a sample, please explore the platform as there could be something perfect for you if not listed here.

Pros & cons

Pros

  • Pay reliability is a standout — almost nobody reports not being paid, and regulars say they're paid on time every time.
  • Open worldwide, so it's accessible where many other platforms aren't.
  • The AI interview process recently got faster, reusing past results instead of re-interviewing for every role.
  • Strong rates and a positive experience once you're actually placed on a project.

Cons

  • Getting in is the hard part — many apply to 20+ roles and get stuck in a 'Hiring Manager Review' / 'Did Not Qualify' loop.
  • Interview feedback can feel unclear and like a moving target.

Back to top ↑

Ethos (AskEthos)

Apply to Ethos →

About

Ethos (AskEthos) is an expert network that onboards you through an AI chat or voice interview and matches you to projects — AI model training and evaluation (including work supporting major labs), data annotation, and expert consulting or advisory calls. Advertised pay ranges roughly $35–$225/hr with a median around $80.

Countries served

Worldwide, subject to ID, sanctions and payment checks — residency can affect eligibility for some projects.

Areas of expertise

This is just a sample, please explore the platform as there could be something perfect for you if not listed here.

Pros & cons

Pros

  • Sign-up is quick and easy, and matching to opportunities can happen fast after the AI interview.
  • Multiple people report legitimate first payouts after about a month, with smooth pay once work clears.
  • A broad and growing set of opportunities across many expert fields.

Cons

  • Payout delays are a recurring complaint, with some reporting being ghosted by support after offboarding.
  • Offboarding over quality flags (including work wrongly flagged as AI-generated) is contested as unfair, and identity or residency checks have ended some engagements with pay left uncertain.

Back to top ↑

DataAnnotation

Apply to DataAnnotation (pick your field) →

About

DataAnnotation.tech hires remote contributors to train and evaluate AI models — writing, code evaluation and debugging, fact-checking, chatbot rating and more. You pass a qualification assessment to get in. Specialist tracks pay roughly $50–$100+/hr; generalist work is closer to $20–$30/hr. Our apply link lets you choose the field that fits you.

Countries served

US, Canada, UK, Ireland, New Zealand and Australia.

Areas of expertise

This is just a sample, please explore the platform as there could be something perfect for you if not listed here.

Pros & cons

Pros

  • It's widely agreed to pay — most contributors are paid reliably, typically weekly.
  • Good hourly rates, with developers reporting around $50–$75/hr.
  • Flexible hours, and work tends to become steadier once you're established, especially for detail-oriented writers.
  • Long-term earners exist — people with 2–3 years and thousands of hours on the platform.

Cons

  • The qualification assessments are long and unpaid, and acceptance is competitive — some feel the vetting is unpaid labor.
  • The most serious recent complaints are sudden account closures with approved earnings (sometimes $1,000+) withheld and little response from support.

Back to top ↑

Surge AI

Visit Surge AI →

About

Surge AI is a premium, largely invitation-only human-data firm — now one of the biggest by revenue — running RLHF, NLP labeling, code generation and search/ad evaluation, with elite consultant tracks in law, medicine, finance and academia. Base pay is around $60/hr, and specialist or fellow tracks are advertised as high as $200–$500/hr. Many contributors actually enter through its DataAnnotation.tech channel.

Countries served

US, UK, Canada, Australia and other English-speaking countries, though individual projects may restrict further.

Areas of expertise

This is just a sample, please explore the platform as there could be something perfect for you if not listed here.

Pros & cons

Pros

  • Widely seen as legitimate, with a reputation for real, higher-quality reasoning and writing work rather than tedious labeling.
  • Very high advertised ceilings on elite tracks ($200–$500/hr for author, medical, legal and finance specialists).
  • A $100-to-apply plus $100-if-accepted sign-up bonus, and an easier back door via DataAnnotation.

Cons

  • No public contributor sign-up — entry points are confusing and the elite track is invite-only.
  • The full-time hiring process draws complaints of large unpaid take-home assignments followed by quick rejections.

Back to top ↑

Snorkel AI

Visit Snorkel AI →

About

Snorkel AI is a Stanford spinout that runs an expert community creating and evaluating data across 1,000+ domains — math, law, finance, coding, the sciences, medicine and humanities. It advertises up to $100/task with a $500/day cap and $3,000+/week for top performers, and usually expects an advanced degree.

Countries served

US only (California currently excluded).

Areas of expertise

This is just a sample, please explore the platform as there could be something perfect for you if not listed here.

Pros & cons

Pros

  • Among the highest published per-task rates, aimed at credentialed (Master's/PhD) experts who face less race-to-the-bottom competition.
  • A clear advancement path from freelance to dedicated or reviewer roles, with speed and quality bonuses.
  • Several contributors call it underrated and genuinely good work.

Cons

  • Recurring payment and offboarding horror stories — e.g. a reviewer terminated with ~$1,000 unpaid and silent support.
  • Assessment and review limbo is common, with submissions stuck 'pending' for weeks with no score.

Back to top ↑

Alignerr

Visit Alignerr →

About

Alignerr is Labelbox's contributor arm for AI training — data labeling, coding and ML (SWE), model alignment and evaluation, and domain-expert annotation. Rates are advertised around $40–$120+/hr, and work typically moves from a (sometimes unpaid) evaluation phase into paid production.

Countries served

US preferred; international payouts available, though many regions are geo-restricted.

Areas of expertise

This is just a sample, please explore the platform as there could be something perfect for you if not listed here.

Pros & cons

Pros

  • Confirmed legit (Labelbox) and reported to pay well and on time once you're in production, with some preferring it over Outlier.
  • Solid advertised rates for generalist and writer roles.
  • An active, informative community that explains the eval-then-production pay structure.

Cons

  • Heavy 'unpaid eval' complaints — pay is often hidden until you pass evals, meetings and lengthy instructions done for free.
  • Payment delays and chasing are reported, and projects can pause or vanish from the dashboard with no visibility.

Back to top ↑

Handshake AI

Visit Handshake AI →

About

Handshake AI is the AI-training arm of the Handshake careers platform, running expert model-evaluation and subject-matter tasks through its Fellowship program. Project rates reach roughly $125/hr, and it's open to students, grads and professionals.

Countries served

US only for the work (valid US work authorization; CPT/OPT may qualify).

Areas of expertise

This is just a sample, please explore the platform as there could be something perfect for you if not listed here.

Pros & cons

Pros

  • Broadly viewed as legit, fully remote and flexible, and open to all career stages with no .edu email required.
  • High earning potential on the right projects — reports of ~$5k/week and $50–$125/hr, sometimes with multiple offers at once.
  • Year-long contributors describe it as worthwhile side income and a good learning experience.

Cons

  • Serious nonpayment and ban allegations exist, including a fellow claiming ~$23,000 owed after a sudden ban.
  • Abrupt offboarding and paused projects are common, so it is unstable as a primary income.

Back to top ↑

Mindrift

Visit Mindrift →

About

Mindrift (by Toloka) is a freelance AI-training platform for STEM, coding (review and generation), multimodal and POV video and domain-expert data creation across 90+ domains. Advertised rates run $60–$100+/hr on expert projects, with entry tasks closer to $15–$30.

Countries served

Global, across 20+ languages (sanctions and geo restrictions apply).

Areas of expertise

This is just a sample, please explore the platform as there could be something perfect for you if not listed here.

Pros & cons

Pros

  • Legitimate with a wide variety of roles and a low barrier to entry, even for those with little experience.
  • Attainable rates of $20+/hr on some projects once established, with Toloka's infrastructure behind it.
  • Flexible, remote, task-based work with an active community for help.

Cons

  • Strong 'free labour' warnings — very low training pay (e.g. 10+ hours for $30–$40).
  • Stuck or locked earnings and sudden suspensions for vague 'anomalies' that freeze withdrawals.

Back to top ↑

Outlier (Scale AI)

Visit Outlier →

About

Outlier (by Scale AI) offers generalist and expert AI-training tasks — response rating, coding, STEM, multilingual and domain-expert work (cardiology, dermatology, finance), plus voice evaluation. Top rates reach ~$100/hr, though general work often sits in the $15–$50 range.

Countries served

~50+ countries, with pay priced by location and language.

Areas of expertise

This is just a sample, please explore the platform as there could be something perfect for you if not listed here.

Pros & cons

Pros

  • Multiple 2026 posts confirm it's legit and pays reliably — one contributor is 'paid every Tuesday like clockwork'.
  • Genuinely lucrative projects surface for skilled coding and technical specialists.
  • Fully remote and flexible, with large project variety when queues are active.

Cons

  • The new 'standard/exceptional' pay system is widely seen as a disguised pay cut.
  • Often called the lowest-paying of the major platforms, with chronic empty-queue droughts.

Back to top ↑

xAI (Human Data / AI Tutor)

Visit xAI careers →

About

xAI hires contributors directly (often W2 or contract) to train Grok — data annotation and tutoring across text, voice and video, spanning STEM, legal and compliance, medicine, humanities and multilingual audio. Pay runs roughly $35–$100/hr, with software-engineering specialists at the higher end.

Countries served

US plus remote international, though some US states are excluded (e.g. Wyoming, Illinois).

Areas of expertise

This is just a sample, please explore the platform as there could be something perfect for you if not listed here.

Pros & cons

Pros

  • Advertised as a stable ~$40/hr, fully remote, full-time role rather than feast-or-famine queues.
  • Direct-hire structure with team-lead support and an emphasis on quality over quantity.
  • Straightforward application — a resume drop plus a short optional cover letter.

Cons

  • Repeatedly described as hard, stressful work — 'not an easy-money job'.
  • Requires constant self-generated creativity for hours, which many find draining.

Back to top ↑

Hatch (by NobleStark)

Visit Hatch →

About

Hatch (by NobleStark) is a curated expert network building data, benchmarks and evaluations for AI labs — software engineering, cybersecurity, math, code and reasoning. It advertises top-of-market rates with weekly payouts and selects for genuine senior experts.

Countries served

Global expert network (5,000+ vetted contributors).

Areas of expertise

This is just a sample, please explore the platform as there could be something perfect for you if not listed here.

Pros & cons

Pros

  • Positioned for genuine senior experts (engineers, PhDs, competitive-programming veterans), not annotation-farm labor.
  • Advertises top-of-market rates with weekly payouts.
  • Selective entry (a short skill task plus a live review with a domain lead) suggests curated, higher-quality work.

Cons

  • Essentially no firsthand contributor reviews on Reddit — claims come mostly from the company’s own marketing.
  • Highly selective vetting means many applicants may never be matched to paid work.

Back to top ↑

Braintrust

Visit Braintrust →

About

Braintrust is a zero-fee, token-based talent network running domain-expert AI training — medicine, law, finance, education and engineering, plus data review. Rates span roughly $20–$200/hr, with project-based rather than micro-task engagements.

Countries served

Global (zero fees to talent).

Areas of expertise

This is just a sample, please explore the platform as there could be something perfect for you if not listed here.

Pros & cons

Pros

  • Zero-fee, decentralized marketplace matching vetted talent to higher-skill AI and engineering roles.
  • Some contributors report transparent processes and real paid work on specific projects.
  • Can act as a gateway into larger programs for matched candidates.

Cons

  • Many users find it scam-feeling, with some reporting shockingly low rates (e.g. $15/hr for work others did at $20).
  • Frequent complaints that no actual project materializes after acceptance.

Back to top ↑

Meridial (by Invisible)

Visit Meridial →

About

Meridial (by Invisible Technologies) posts hundreds of freelance, domain-expert AI projects at once — law, STEM, finance, coding, linguistics and AI safety. Reported rates span roughly $6–$150/hr, and it pays contributors on a regular (roughly fortnightly) cycle.

Countries served

Global (contractor; rates confirmed per project).

Areas of expertise

This is just a sample, please explore the platform as there could be something perfect for you if not listed here.

Pros & cons

Pros

  • Confirmed legit (run by Invisible) with firsthand reports of reliable pay every two weeks.
  • Pays for set onboarding hours, unlike many platforms that onboard unpaid.
  • Posts a large volume of projects across a wide pay range, and is seen as relatively transparent.

Cons

  • Projects are often small and short-lived, limiting earning potential.
  • Hard to land work — some never pass scoring or hit verification failures (e.g. native-language checks).

Back to top ↑

Pareto AI

Visit Pareto AI →

About

Pareto AI is a talent-first human-data platform built around referral sourcing — expert labeling, reasoning, LLM-debate judging and fine-tuning data. Rates are around $35–$50/hr, focused on higher-skill reasoning work rather than microtasks.

Countries served

Global, with a large contributor base in Kenya.

Areas of expertise

This is just a sample, please explore the platform as there could be something perfect for you if not listed here.

Pros & cons

Pros

  • Contributors report attractive expert-project rates of roughly $35–$50/hr.
  • Appears legitimate, with some applicants hearing back within a few days.
  • Work centers on higher-skill reasoning and fine-tuning data.

Cons

  • Pay rate is often withheld until after onboarding, with recruiters deflecting direct questions.
  • Described as overcrowded — many apply repeatedly and never hear back.

Back to top ↑

iMerit Scholars

Visit iMerit Scholars →

About

iMerit Scholars is iMerit’s elite program for domain-expert GenAI tuning and evaluation across specialized fields, structured as longer-term enterprise projects rather than one-off gigs. Pay is set by location, experience and skill.

Countries served

Global (rate set by location, experience and skill).

Areas of expertise

This is just a sample, please explore the platform as there could be something perfect for you if not listed here.

Pros & cons

Pros

  • Positioned as structured, long-term enterprise projects rather than one-off gigs.
  • Early-stage communication and onboarding are reported as responsive, with assessments easy to pass.

Cons

  • Multiple recent posts allege sudden mid-task offboarding with withheld pay (one offboarded at 3.9 of 4 required hours).

Back to top ↑

Turing

Visit Turing →

About

Turing runs an AI-training data track (code writing, evaluation and validation, reasoning, STEM, multilingual and agentic data) alongside its US developer-placement business. AI-training pay runs roughly $15–$75/hr, and its Frontier Contributor Program is a paid research-data track for researchers.

Countries served

Global (the AI-lab track is open internationally).

Areas of expertise

This is just a sample, please explore the platform as there could be something perfect for you if not listed here.

Pros & cons

Pros

  • Some contributors report smooth workflow and on-time payments, and the Frontier Contributor Program is a legitimate paid research track.
  • Rates can be favorable for contributors in the global south thanks to USD conversion.
  • Team culture is described by several as supportive and collaborative.

Cons

  • A detailed recent account describes chaotic onboarding, unclear payments and broken tooling on one project.
  • Pay is inconsistent and vendor-dependent — direct hires and third-party hires get very different rates.

Back to top ↑

CrowdGen (by Appen)

Visit CrowdGen →

About

CrowdGen (by Appen) runs multilingual general LLM training plus domain-specific tasks — math, physics, chemistry, medicine, nursing and philosophy — drawing on Appen’s global crowd. Domain tracks pay higher than generalist work.

Countries served

Global (170+ countries via Appen).

Areas of expertise

This is just a sample, please explore the platform as there could be something perfect for you if not listed here.

Pros & cons

Pros

  • Backed by an established company (Appen), with a real contributor community and active earnings megathreads.
  • After a public payment crisis, Appen’s CEO acknowledged the failures and some contributors confirmed eventually being paid.

Cons

  • The 2024 migration from Appen Connect to CrowdGen triggered widespread payment failures, with some never paid for completed work.

Back to top ↑

OneForma (by Centific)

Visit OneForma →

About

OneForma (by Centific) offers per-task data work — online rating, data collection and annotation across text, image, audio and video, search and ad evaluation, and health-data review. Some submissions pay in the $60–$150 range, and payouts typically land the 10th–20th of the following month.

Countries served

Global community of experts.

Areas of expertise

This is just a sample, please explore the platform as there could be something perfect for you if not listed here.

Pros & cons

Pros

  • Legitimate per-task data and ad-evaluation work is available, and the platform does pay many contributors.
  • An active official subreddit shares project and payment guidance.

Cons

  • The dominant 2026 complaint is purchase orders undercounting tracked hours (e.g. 33 hours paid as 6.5).

Back to top ↑

Welo Data (Welocalize)

Visit Welo Data →

About

Welo Data (by Welocalize) runs multilingual AI evaluation, safety testing, data-quality assessment and image 'golden answer' annotation, drawing on a huge pool of vetted experts. Typical rates are modest (~$19–$26/hr), though some gigs are quoted higher.

Countries served

Global (500,000+ vetted experts).

Areas of expertise

This is just a sample, please explore the platform as there could be something perfect for you if not listed here.

Pros & cons

Pros

  • Regarded as a legitimate, established language-services company that generally pays contributors.
  • Flexible remote work with some attractive quoted rates (e.g. ~$40/hr on certain evaluation gigs).
  • Appeals to linguists, and the work itself is described as easy when tasks are available.

Cons

  • Payment and invoicing are a recurring complaint, including a recent reviewer stuck unpaid over missing POs and unresponsive support.
  • Task availability is inconsistent, with raters reporting no tasks for weeks.

Back to top ↑

Sigma AI

Visit Sigma AI →

About

Sigma AI runs multilingual data collection and annotation — voice and audio collection, transcription and labeling across modalities — in hundreds of languages. Third-party reports cite a pay band of roughly $19–$51/hr.

Countries served

~99 countries and 700+ languages (native speakers).

Areas of expertise

This is just a sample, please explore the platform as there could be something perfect for you if not listed here.

Pros & cons

Pros

  • Multiple users confirm it is legitimate and actually pays, with projects recurring month-to-month.
  • Genuine flexibility and a low-barrier entry via transcription and reasoning tests in your native language.
  • Varied multilingual work across many languages, with a fairly wide reported pay band.

Cons

  • Workload can be heavy and draining — one worker cited quotas of ~450–500 short audio clips per hour for low pay.
  • Communication gaps are common, including signed contracts where the project never starts.

Back to top ↑

Remotasks (Scale AI)

Visit Remotasks →

About

Remotasks (by Scale AI) specializes in image, video and LiDAR/3D annotation and multimodal labeling. Generalist pay is low ($3–$15/hr) with specialist work around $15–$50, and qualification exams unlock better-paying tasks.

Countries served

Global (availability is geographically volatile).

Areas of expertise

This is just a sample, please explore the platform as there could be something perfect for you if not listed here.

Pros & cons

Pros

  • A real platform that does pay, especially well-suited to image, video and LiDAR computer-vision annotation.
  • Structured training courses and qualification exams can unlock better-paying specialized work.
  • Open to a global pool with large task variety.

Cons

  • Basic tasks have largely dried up, and mandatory unpaid training drags effective pay to roughly $3–$10/hr.
  • Long dry periods with empty queues make income inconsistent.

Back to top ↑

Toloka

Visit Toloka →

About

Toloka runs annotation, labeling, AI-tutor dataset creation, evaluation and multimodal tasks, with flexible PayPal/Payoneer withdrawals after approval. Average pay is around $44/hr on better projects, though many tasks pay far less.

Countries served

Global (sanctions and geo restrictions apply).

Areas of expertise

This is just a sample, please explore the platform as there could be something perfect for you if not listed here.

Pros & cons

Pros

  • Repeatedly described as legitimate and genuinely paying, with flexible withdrawals anytime after approval.
  • Can be good quick money for consistent workers, especially on video and recording tasks.
  • Tasks are generally simple and low-barrier to start.

Cons

  • Pay is frequently called 'peanuts', with many non-video tasks paying very little.
  • Slow QA causes long payment delays and pending balances (one user reported ~$5k stuck).

Back to top ↑