Joining TrainAI is free, with no prior training or experience required and a minimum age of 18. For tasks, contributors supply their own Windows/macOS computer, internet, email and current Chrome browser.
Age and experience describe community entry; equipment is required to do tasks. Projects have separate training and tests.
RWS describes language, general machine-learning and project-specific tests. It says contributors must pass several project training rounds before starting work.
The client-facing FAQ describes project preparation. Training pay is not stated.
RWS's TrainAI community FAQ says it pays per completed task and communicates project-specific payment details in advance. It lists PayPal or direct bank payment and lets contributors select their preferred payment method during registration.
Community FAQ payment terms, not observed payout results or a numerical rate. Task payments vary by project. The page does not establish a currency, payment schedule, payout or withdrawal fees, third-party fees, minimum withdrawal, or PayPal access for a particular country or account. Bank payment is not identified as ACH. Payment support does not establish residence eligibility, current work availability or guaranteed earnings or receipt. Retrieval and review dates do not establish a policy change or validity period.
RWS's Japanese Search Engine Evaluator listing offers payment through PayPal or a bank account.
This temporary, remote contract role requires current residence in Japan, Japanese proficiency and English at B2 level or higher; the payment methods are scoped to that listing.
RWS advertises up to USD 15/hour for a Japanese Search Engine Evaluator role, described on the page as approximately JPY 2,300/hour.
One role for Japan residents proficient in Japanese and English. Training modules and a client test are required. This is an advertised maximum, and the yen equivalence is RWS’s own approximation; it is not a verified exchange-rate conversion.
Project location and language requirements are separate
Official information
RWS's Japanese Search Engine Evaluator listing requires current residence in Japan, knowledge of local culture, Japanese proficiency and at least B2 English.
One TrainAI role; not a statement that all Japanese-language work requires residence in Japan or that every Japan resident qualifies.
An Italy applicant reported no tasks after completing setup
Worker report
A Reddit poster who said they applied from Italy for Speech AI Evaluation Specialist reported completing account setup and tests but having no tasks the following day.
One applicant's early experience, with an unverified publication date. It does not show how long they eventually waited or establish a restriction on Italy.
RWS describes community members as freelancers or private individuals who can choose when to work, with the relationship governed by its terms and an NDA.
This general FAQ wording does not establish the schedule for a particular assignment.
RWS says TrainAI rates consider in-country minimum wages, average salaries, cost of living and inflation, alongside skills and the contributor's desired pay.
The FAQ describes how rates are set but provides no country rate table.
Selected public accounts. Experiences are self-reported.
Historical payment follow-up
Undated
Contributors in a 2025 discussion reported pending project payments. One later updated that payment had arrived; another said their other projects had paid while Ocean remained outstanding.
Separate contributor accounts naming TrainAI projects; discussion includes November and December 2025 payment references.
Loekalization’s November 2025 blog reported AI contractors’ accounts of delayed RWS payments, revisited in March 2026. RWS’s own statement acknowledged delays on some TrainAI projects and attributed them to payment-system issues.
An independent localization business’s commentary and RWS’s response concerning 2025 delays.
Acolad’s freelance talent pool for AI data collection, annotation, language work and content validation, with contributors contacted for matching projects.
Data annotation · Data collection · Language & writing · AI evaluation
Lionbridge’s contributor community for AI response evaluation, annotation, recordings and language work, with participants invited to matched projects.
AI evaluation · Data annotation · Data collection · Language & writing