TrainerScout

RWS TrainAI

RWS's AI-data community: language, search-evaluation and annotation work in 175+ countries.

Last reviewed 2026-10-14

See RWS TrainAI's open roles

Typical pay

Search evaluators up to $5–30/hr by country; legal annotators $16.20–36.45/hr

Who can join

Native speaker of the target language plus English, living in the posted country, 18+

How you get in

Sign-up form, recruiter contact, then a client qualification test

Payouts

PayPal or bank account; schedule not published

Watch out

Payment-delay complaints (Trustpilot 1.8/5); some country rates are very low

What it is

  • Official

    TrainAI is the AI training-data business of RWS Holdings plc, a UK language-services and technology group listed on London's AIM market. RWS launched TrainAI as a branded service in 2023.

    rws.com· checked 2026-10-14
  • Official

    RWS describes the TrainAI community as 100,000+ annotators and linguists working in 400+ language variants across 175+ countries.

    rws.com· checked 2026-10-14
  • Official

    Work is done through the TrainAI community, which RWS recruits for mainly through its public Lever board. All 94 live postings on that board on 2026-10-14 were in the TrainAI team, and all were remote.

    api.lever.co· checked 2026-10-14
  • News

    Business size: in its half-year to 31 March 2026, RWS reported revenue of £360.3m (up about 5%) and credited an "exceptional performance" in TrainAI. AI-related products and services rose to 32% of group revenue.

    investegate.co.uk· checked 2026-10-14
  • News

    On results day (11 June 2026) the shares fell about 15% after RWS cut its interim dividend by 29% and warned of a currency hit to full-year profit. This is a company-level fact, not a TrainAI worker issue.

    ajbell.co.uk· checked 2026-10-14

Coding work

  • Official

    Coding: little or none posted. None of the 94 live TrainAI postings on 2026-10-14 was a coding or software-engineering role. The general "Join our global TrainAI talent pool" listing mentions machine learning and research backgrounds but names no coding project and no pay.

    jobs.lever.co· checked 2026-10-14
  • Official

    Search engine evaluation (largest group, about 40 postings): rate how a client's search engine answers everyday searches, plus text, audio, image and video tasks. 5 to 20 hours per week (one posting gives an average of 15). No previous experience required. Posted pay is "up to" a country rate: from up to $5/hr (Bengali, Marathi, Tamil, Telugu in India) and $6/hr (Mexico, Nigeria, Ukraine, Vietnam) to up to $21/hr (Australia), $22/hr (Denmark), $23/hr (Norway) and $30/hr (Switzerland). France and Germany are quoted in euros (up to €13/hr and €14/hr).

    jobs.lever.coapi.lever.co· checked 2026-10-14
  • Official

    AI Data Specialist (language): data collection, evaluation and annotation, pairwise comparisons, counting tasks and object tagging across audio, video and images. Freelance, 10+ hours per week, flexible. Posted rates: $16.50/hr (France), $18/hr (Germany, Japan), $11/hr (Italy), $10/hr (Korea), $8/hr (Vietnamese posting). Each says "rates vary per country".

    jobs.lever.co· checked 2026-10-14
  • Official

    Legal annotators (legal): contract and clause analysis for contract-management AI. Freelance, 25+ flexible hours per week. Requires contract or legal-domain expertise plus fluency in English and the target language. Posted rates run from $16.20/hr (Chinese Simplified; one Russian posting) to $36.45/hr (Dutch), with Korean and Portuguese (Portugal) at $22.95/hr.

    jobs.lever.coapi.lever.co· checked 2026-10-14
Show 4 more
  • Official

    Brazilian Portuguese subject-matter experts (writing / general knowledge): 14 postings by topic (for example history, law and ethics, sports, arts, food). The work is writing in-depth answers and questions that test AI reasoning. Use of AI tools to write them is "strictly prohibited". Part-time, 10+ hours per week. Posted rates: $21.60/hr (several topics), $26.40/hr, $30/hr (History, Religion & Philosophy) and $36/hr (Law & Ethics).

    jobs.lever.coapi.lever.co· checked 2026-10-14
  • Official

    Speech and audio evaluation (language): Speech AI Evaluation Specialist postings list $10/hr to $18/hr. Generative Audio Evaluation (lip-sync, caption compliance, voice quality) lists $19/hr (Dutch) and $11/hr (Polish), with a test due within 48 hours of assignment. AI Linguistic Evaluator postings (Marathi, Telugu) list $16/hr and $18/hr for a project of about 10 to 15 hours in total. AI Conversation Specialist (Saudi Arabic dialects) lists $15/hr.

    jobs.lever.cojobs.lever.co· checked 2026-10-14
  • Official

    US part-time W-2 annotators: five "General AI Data Annotator" postings (French, German, Italian, Japanese, Korean plus English) are part-time W-2 employee roles at $21.50/hr, 20 hours per week (4 hours a day, Monday to Friday), with up to 9 extra hours a week based on demand and paid training. Open only to residents of listed US states (for example CA outside the SF Bay Area, NY outside New York City, TX, WA).

    jobs.lever.co· checked 2026-10-14
  • Official

    Contract type for everything else: freelance. Applicants choose "Freelancer" (has a business licence, VAT number or EIN) or "Private Individual". RWS says community members "are not bound by a contract" and can work whenever they like, though work is governed by terms and conditions and an NDA.

    jobs.lever.corws.com· checked 2026-10-14

Eligibility

  • Official

    Countries: recruiting is country by country. Postings name a country of residence (for example, "Currently reside in Australia"), and the pay rate depends on that country. RWS says the community spans 175+ countries.

    jobs.lever.corws.com· checked 2026-10-14
  • Official

    Language: most roles need native-level fluency in the target language plus English (B2 to C2 for AI Data Specialist roles).

    jobs.lever.co· checked 2026-10-14
  • Official

    Search evaluator conditions: 18+; one Search Quality Rater per household; must sign the project NDA; needs a smartphone, a personal computer, reliable internet and antivirus software; must pass training modules and a client-made test before starting.

    jobs.lever.co· checked 2026-10-14
  • Official

    Credentials for expert roles: legal annotator roles ask for contract or legal-domain expertise; Brazilian SME roles ask for academic or industry expertise; the AI Linguistic Evaluator roles ask for a degree in linguistics, translation or a related field (or equivalent).

    jobs.lever.cojobs.lever.co· checked 2026-10-14
  • Official

    No fee: RWS says joining the TrainAI community is free and it will never ask applicants for payment.

    rws.com· checked 2026-10-14

How to get in (format only)

  • Official

    Steps: choose a vendor type (Private Individual, Sole Proprietor / Freelancer, or Company), complete the sign-up form, and the recruiting team contacts you about next steps. Questions go to AICommunity@rws.com.

    community.rws.com· checked 2026-10-14
  • Official

    Talent pool: the "Future Opportunities" listing says RWS will reach out when a matching opportunity becomes available. It gives no timeline.

    jobs.lever.co· checked 2026-10-14
  • Official

    Tests: search evaluator roles require training modules and a client-made qualification test. Audio roles list language knowledge assessments, internal testing and project-specific tests (for example TTS or transcription testing).

    jobs.lever.cojobs.lever.co· checked 2026-10-14

Pay and payments

  • Official

    Rate setting: RWS says rates are based on economic factors plus your desired rate and skillset, with annual reviews. A community FAQ lists minimum wage, average salary, cost of living and inflation in the country among the factors.

    rws.comcommunity.rws.com· checked 2026-10-14
  • Official

    Per task or per hour: the community FAQ says payment is per completed task with details given in advance. Most current postings quote an hourly rate.

    community.rws.comapi.lever.co· checked 2026-10-14
  • Official

    Method: PayPal or bank account, chosen at registration. Search evaluator postings say "Payments via PayPal/Bank Account".

    rws.comjobs.lever.co· checked 2026-10-14
  • Worker reports

    Posts on RWS's own community forum say payouts run through Tipalti, that Tipalti does not support PayPal in some countries (one poster names Nigeria), and that some setup emails never arrived.

    community.rws.com· checked 2026-10-14
Show 2 more
  • Official

    Schedule: no official payout schedule was found.

    rws.com· checked 2026-10-14
  • Worker reports

    Reported delays: recent Trustpilot reviews (May to Oct 2026) describe late or missing payments, missing work orders and slow support, including one report of payments taking "2-3 months".

    trustpilot.com· checked 2026-10-14

Work availability

  • Official

    Official position: postings list start date "Immediately" and duration "TBC". Some are fixed-size projects (for example about 10 to 15 total hours for a linguistic evaluation). Search evaluator roles set 5 to 20 hours per week.

    jobs.lever.cojobs.lever.co· checked 2026-10-14
  • Worker reports

    Review blogs describe project-based work with gaps between projects and empty queues during transitions. One rates work consistency 2 out of 5.

  • Worker reports

    One Trustpilot reviewer (May 2026) describes taking a test and doing some work, then hearing nothing further.

    trustpilot.com· checked 2026-10-14

Risks and controversies

  • Worker reports

    Payment complaints: RWS's Trustpilot profile is 1.8/5 from 53 reviews, 87% 1-star. Payment is the most common complaint in recent reviews. The profile is unclaimed and RWS has not invited reviews, so the sample is small and self-selected.

    trustpilot.com· checked 2026-10-14
  • Worker reports

    Forum payment threads: RWS's own community forum has threads titled "Anyone got payments ever???" and "Payment Issue", plus a reported failed USD 87.50 payment due to bank-detail rejection.

  • Official

    Low rates in some countries: posted search evaluator rates are up to $5/hr in several Indian languages and up to $6/hr in Mexico, Nigeria, Ukraine and Vietnam. These are maximums ("up to").

    api.lever.co· checked 2026-10-14
  • News

    Company finances: RWS cut its interim dividend by 29% in June 2026 and net debt rose to £32.5m. No link to contractor payments was reported.

    ajbell.co.uk· checked 2026-10-14

Referral program

  • None found. No referral program is described on RWS's TrainAI community page or in the live postings, and a third-party platform directory lists "No referral or affiliate program found".

    rws.comaitrainer.work· checked 2026-10-14

Worker sentiment (reports, not verified)

  • Ratings: Trustpilot (rws.com) 1.8/5 from 53 reviews, 32 in the last 12 months. The profile covers all of RWS, not only TrainAI.

    trustpilot.com· checked 2026-10-14
  • Negative, top theme: late, partial or missing payments and slow support. Reported on Trustpilot (2026) and summarized by an independent review blog (Sept 2026) as "a serious pattern of recent payment and support complaints".

  • Negative: unpredictable project flow and silence after testing.

    selfmadesuccess.com· checked 2026-10-14
  • Positive: flexible hours, low entry barrier for general and search roles, and "fairly good" pay in one 3-star review that criticized management.

    trustpilot.com· checked 2026-10-14
  • Rough balance: public reviews skew negative on payment reliability; positives center on flexibility and accessibility.

    trustpilot.com· checked 2026-10-14

Quick answers

How much does RWS TrainAI pay?
Search evaluators up to $5–30/hr by country; legal annotators $16.20–36.45/hr
Who can join RWS TrainAI?
Native speaker of the target language plus English, living in the posted country, 18+
How do you get into RWS TrainAI?
Sign-up form, recruiter contact, then a client qualification test
How and when does RWS TrainAI pay out?
PayPal or bank account; schedule not published
What should you watch out for at RWS TrainAI?
Payment-delay complaints (Trustpilot 1.8/5); some country rates are very low