An English evaluator in the Philippines, a Rust developer working from almost anywhere, and a film professional judging AI-generated video might all be described as AI trainers. Their work, eligibility rules, and advertised pay are nothing alike.
That is the useful part of this month’s search. Instead of another list of companies that might recruit, I checked ten identifiable role pages across Mercor, micro1, Ethos, Outlier, and Turing.
They span education, film, coding, document quality assurance, infrastructure, multimedia evaluation, and video annotation. New to the field? Our introduction to AI training explains how human contributors fit into the process.
A date matters here. These listings were publicly discoverable during the November 5, 2026 check. That does not confirm unfilled vacancies, guarantee a place on a project, or mean an applicant in every country can qualify. Check the linked role page again immediately before applying. Advertised pay is not verified take-home pay.
The 10 opportunities at a glance
| Platform | Role | Advertised compensation | Important catch |
|---|---|---|---|
| Mercor | Education & Training Professional | $30/hour | 3+ years’ experience; about 10 hours/week; location restriction |
| Mercor | Film & Video Craft Expert | $70–$100/hour | Film or VFX credentials; 20+ hours/week; employment terms need clarification |
| micro1 | Rust Architect/Backend Developer | $50–$70/hour advertised | Output-based payment also stated; approximately 15 hours/week |
| micro1 | Software Engineer, Open Source Contributor | $50–$100/hour advertised | Strong open-source evidence or technical screening; task-based workflow |
| Ethos | Document, Deck & Spreadsheet QA Specialist | $70/hour | 5+ years reviewing professional business documents |
| Ethos | Incident Management/SRE & Reliability Lead | $80/hour | 4+ years owning production incidents and reviews |
| Outlier | AI Image and Video Judge | Up to $18/hour headline | Core work described as up to $15/hour; location-dependent |
| Outlier | English-Speaking AI Image and Video Evaluator | Up to $7/hour | Philippines-specific listing |
| Turing | LLM Trainer, Video Annotation Specialist | Not publicly stated | Eligible-country list; minimum 20 hours/week and Pacific Time overlap |
| Turing | Senior Code Annotator, JS/TS | Not publicly stated | Experienced programmer; regional limits and live coding interview |
The table reports the listing language, not an independently verified offer or ongoing workload. Turing’s public role pages do not state a pay figure for the two roles above, so no amount is estimated.
Mercor: Education work and a surprisingly specialized video role
The first Mercor opening is aimed at teachers, tutors, instructional designers, curriculum developers and learning-and-development professionals with at least three years’ experience. At $30/hour for approximately ten hours weekly, the assignment involves giving an AI assistant realistic educational tasks, working through several conversational turns, and judging the deliverable against professional standards. Its wording restricts eligibility to countries where the assistant is available, so “remote” does not mean every country qualifies.
The Film & Video Craft Expert role is quite different. It advertises $70–$100/hour for people with real experience in filmmaking, cinematography, editing, visual effects or performance. Contributors assess AI-generated video for framing, lighting, continuity, realism and execution of the prompt. The role asks for reliable weekday availability of at least 20 hours.
One important ambiguity: that film listing describes employment through Cincinnatus LLC in one section but includes generic independent-contractor language elsewhere. Applicants should ask which legal arrangement, benefits and payment terms apply to their country before accepting an offer. Don’t assume the same contract conditions as a standard Mercor project.
If you are new to the company, our Mercor review explains its AI interview, project selection and weekly-payment system. You can also join Mercor through this link to explore other roles suited to your experience.
micro1: Attractive coding rates, but read the payment unit
The Rust Architect/Backend Developer listing advertises $50-$70/hour, describes the work as globally remote, and suggests approximately 15 hours per week. The tasks involve production Rust, concurrency, API behavior, debugging and the design of reproducible backend tests.
There is a catch right inside the role description. Despite the hourly headline, the compensation section states that experts are paid per task meeting project specifications, with minimum submission requirements. The advertised range therefore should not be treated as a guaranteed paid hourly rate. Ask how task prices, expected time, revisions and rejection decisions work.
Another micro1 software engineering listing advertises $50-$100/hour for open-source contributors who can build reproducible reinforcement-learning environments for AI coding models. It mentions approximately 15 hours per task run. Candidates should be ready to demonstrate substantive GitHub contributions or complete a technical assessment.
Both roles illustrate why the headline number is only half the story. Read our breakdown of what AI training jobs actually pay before comparing a per-task assignment with a time-based contract. Explore micro1’s expert opportunities if coding is your strength.
Ethos: Professional judgment is the qualification
Two Ethos listings show how far AI evaluation has moved beyond basic response ranking.
The Document, Deck & Spreadsheet QA Specialist advertises $70/hour, generally 5–20 hours weekly, for applicants with at least five years reviewing business-ready documents. The work resembles a demanding final quality check: spotting incorrect formulas, inconsistent figures, broken references, formatting problems and discrepancies across presentations and spreadsheets. You need production-level skill, not simply familiarity with office software.
The Incident Management/SRE & Reliability Lead advertises $80/hour and a similar flexible weekly commitment. Applicants need at least four years in production reliability or incident leadership. Tasks involve examining AI-generated postmortems, escalation runbooks, root-cause analyses, SLO reports and remediation trackers.
Both pages describe an AI voice screening stage and say work can begin as soon as practical. They do not establish that every qualified person will receive a project, or that every country is eligible. Applicants should confirm location and payout support with Ethos before investing in a lengthy application.
Outlier: The same kind of evaluation, very different pay by location
Outlier currently displays multimedia evaluation roles. These involve comparing generated images or videos, identifying visible defects, or judging the quality of AI audio and explanations.
The AI Image and Video Judge listing carries an up-to-$18/hour headline across a published list of eligible countries including India. However, the detailed compensation text says core project work pays up to $15/hour, with potential additional incentives.
That distinction matters: an incentive-inclusive headline is not the same as the base project rate, and onboarding or other non-core work may be paid differently.
A separate English-Speaking AI Image and Video Evaluator listing is explicitly remote within the Philippines and advertises up to $7/hour. It focuses on spotting errors such as distorted anatomy, awkward motion and text mistakes, then writing clear comparisons in English.
The two listings shouldn’t be flattened into a single worldwide pay estimate. They are a good reminder to check the actual country shown on the opportunity. A higher rate on another country’s listing is not your offer.
For a broader overview of checking platform eligibility, see our guide to finding legitimate remote AI training jobs.
Turing: International reach with a real time-zone commitment
Turing’s LLM Trainer, Video Annotation Specialist asks contributors to watch short videos of robots carrying out tasks, identify what happened, and create timestamped annotations of actions and errors.
The listing names India, Pakistan, Nigeria, Kenya, Egypt, Ghana, Bangladesh, Turkey, Brazil and Mexico among its eligible locations. It requires at least 20 hours per week, including four hours of overlap with Pacific Time, and mentions two short sample annotation exercises during screening.
The Senior Code Annotator, JS/TS lists the same countries and weekly minimum. It seeks developers comfortable reading complex code, reviewing generated code and working in AI-assisted development environments. A live technical interview is part of the evaluation.
Neither of these two public pages specifies a numeric rate. That is a reason to ask for the rate and billing rules, not to guess based on other Turing projects. Also check what Pacific Time overlap means for your actual local schedule; a nominally remote contract can still require inconvenient working hours.
Explore Turing’s talent opportunities and use its exact role descriptions to decide whether the workload suits you.
What stood out in this November check
Higher rates usually correspond to specialized judgment. Mercor’s film opening, Ethos’s reliability work and micro1’s advanced coding projects are not interchangeable with generalist annotation. The requirement is often demonstrated professional experience, not a certificate saying you know AI. We explore that shift in our article on why AI companies recruit domain experts.
Not every visible listing is open. During this research, a Mercor English AI Safety Specialist result still appeared in search, but its application page explicitly said it was no longer accepting applications. Mercor redirected interested applicants to a generalist talent network, which is a standing pool rather than a confirmed paid opening. That distinction is worth checking whenever a listing has been circulating for weeks.
Pay models need careful comparison. An advertised $70/hour on a well-defined contract is not the same as a $70/hour equivalent advertised beside per-task payment. And “up to” figures can include conditions or incentives. If you spend time onboarding and waiting between tasks, your effective earnings may be lower than the headline suggests.
Apply selectively, not everywhere at once
Start with the role whose requirements resemble work you have actually done. A teacher can describe real classroom assessment. An editor can explain how they catch document inconsistencies. A backend developer can point to code they designed and debugged. That evidence matters more than generic statements about enthusiasm for AI.
Before applying, check the country, specific role page, rate unit, expected paid workload, screening conditions and payment provider. Keep a copy of the applicable terms, but never retain confidential client material. If assessments lead nowhere, our article on why applicants don’t get selected for AI training jobs discusses possible reasons and ways to improve your approach.
You can compare more platforms in our best AI training platforms guide, or understand the difference between this work and routine labeling in AI training versus data annotation.
The takeaway from these ten listings is straightforward: the best opportunity is not necessarily the one with the biggest advertised number. It is the one you are eligible for, qualified to perform, and able to verify on acceptable payment terms.
Research note: Checked November 5, 2026, using publicly accessible company listings. Vacancy status, country eligibility, headline compensation and assignment availability can change without notice. Listing visibility is not confirmation of a guaranteed opening or contract.
Disclosure: This article contains referral links. I may earn a commission when eligible applicants complete qualifying activities through certain platforms, at no additional cost to them. Referral links do not guarantee hiring, assignments or earnings.
Leave a Reply