Remote AI training jobs for designers
Get paid to critique AI's design work
AI labs pay experienced designers $15 to $350 an hour to score AI-generated interfaces and write grading rubrics. These AI trainer and design evaluator roles are fully remote, freelance, and fill fast. Get an email when new ones open.
- 1Contrast · fail
Billing terms sit at about 2.3:1. They need 4.5:1 to pass WCAG AA. - 2Layout · 4/5
Plans compare cleanly side by side. - 3Hierarchy · 2/5
Both buttons carry equal weight, so there is no primary action.
Current openings
Remote AI training jobs for designers and creatives
We try to keep this list as up to date as possible, but roles fill up quickly, so we recommend applying early. Subscribe to the mailing list to be alerted when new roles are posted.
| Role | Pay | Hours | Location | |
|---|---|---|---|---|
| Up to $320/task | Per task | Remote | Apply | |
| $150–$350/hr | Hourly contract | Remote | Apply | |
| $90–$175/hr | Hourly contract | Remote | Apply | |
| $100–$170/hr | Hourly contract | Remote | Apply | |
| $60–$120/hr | 10+ hrs/wk | Remote | Apply | |
| $80–$110/hr | Full-time | Remote | Apply | |
| $35–$105/hr | Hourly contract | Remote | Apply | |
| $35–$105/hr | Hourly contract | Remote | Apply | |
| $50–$90/hr | 10+ hrs/wk | Remote | Apply | |
| $60–$80/hr | 15–30 hrs/wk | Remote | Apply | |
| $60–$80/hr | Part-time | Remote | Apply | |
| $40–$80/hr | 10+ hrs/wk | Remote | Apply | |
| $15–$80/hr | Hourly contract | Remote | Apply | |
| $21–$70/hr | Hourly contract | Remote | Apply | |
| $20–$50/hr | Hourly contract | Remote | Apply | |
| $21–$35/hr | Hourly contract | Remote | Apply | |
| $21–$35/hr | Hourly contract | Remote | Apply |
See all jobs with pay ranges and eligibility →
Last updated October 11, 2026.
What the work actually is
What AI design evaluator jobs involve
Search for "AI jobs for UX designers" and you mostly get regular design roles that use AI tools. This is different work. AI companies need people with real design judgment to teach models what good design looks like. This is AI evaluation work. Nearly every listing we track asks for one of these.
Grading AI-generated interfaces
Score generated screens and flows on layout, visual hierarchy, typography, and accessibility. Then write a short justification for each score.
Seen at DataAnnotation, Mercor, micro1Writing rubrics
Define what "great" looks like for a design task, so other reviewers and models can be graded consistently. Senior roles lean heavily on this.
Seen at MercorWriting briefs with reference answers
Turn your past projects into realistic design problems, along with a reference answer the model's output gets compared against.
Seen at MercorWritten critique and analysis
Write design rationales, trade-off analyses, research syntheses, and WCAG audits. The writing itself becomes training material.
Seen at micro1, DataAnnotationAnnotating designs into structured data
Break finished layouts into labeled layers, describing hierarchy, grid, color, and type in precise design vocabulary, so models learn how designs are built.
Seen at micro1Curating and rating visuals
Pick the strongest images from a set and explain why: composition, lighting, color, originality. Taste is the skill being paid for.
Seen at micro1Want to try the work before you apply? Run a prompt and a design through the AI Design Evaluation Toolkit, a free tool I built to decode a design prompt and score the result against it.
How hiring works
How to get hired as an AI trainer
Each platform screens differently. On all of them, getting accepted and getting paid work are separate steps, so it's normal to apply to more than one.
Mercor
- Upload your resume and confirm where you work from
- Do a short AI video interview, about 20 minutes
- Get verified and added to the design pool
- Get “instant offers” in your inbox when matching roles open, even for roles you never applied to
Those offers can arrive months after you join, so getting into the system once matters more than chasing every listing. Projects can be extended, shortened, or ended early.
micro1
- Apply to a specific role
- Complete the AI interview
- Meet the minimum criteria and get certified
- Get matched to a project
Certified designers can wait a while for a match if no project in their area is hiring.
DataAnnotation
- Apply on the designer track
- Pass the qualification assessment
- Choose tasks and work flexible hours
Pay depends on task difficulty, and higher-paying work opens up as you prove quality.
Pay rates
How much AI trainer jobs pay designers
Most AI training jobs for designers pay $35 to $110 an hour. Annotation and labeling work pays less, and senior or specialized roles pay much more.
Annotation and labeling
Design critique and evaluation
Senior and specialized roles
Hourly rate (USD)
From Sam, who's been through it
What I wish I'd known
The AI interviews are honestly weird. I took them a few times just to practice before I got comfortable, so don't treat your first one as your only shot. Once I'd done a few and knew what to expect, I could rehearse my answers and sound more eloquent and concise. It helps to write sticky-note reminders and practice STAR answers to a few key behavioral questions, like “Tell me about a time you disagreed with a stakeholder,” so you're not stuck thinking of an answer on the spot. A bonus: after doing a number of these AI interviews, I felt more prepared for in-person interviews too.
Getting placed on a project can take a while. I got an offer months after interviewing, so keep your profile current and don't assume silence means no. On Mercor specifically, once you're in the system, job offers arrive as “instant offers” in your inbox when relevant roles open, often months later and whether or not you applied to them.
Start with generalist roles, and keep an eye out for oddly specific ones, like translators for a particular language or people with experience in a certain business sector. The platforms post new things all the time.
Most of this work is async. You work at your own pace, whenever you want, as long as tasks are available. It's self-directed, with little feedback and almost no meetings or overhead. The nice part is that it's truly remote: most contracts let you work from anywhere, even outside the US.
You'll often be one of a crowd doing the same task, anywhere from 50 to 10,000 people. You won't get the close collaboration of a design team for your tasks (beyond occasional feedback and onboarding instructions), but there are opportunities to ask questions and chat with the other designers on Slack.
This work suits self-directed people who learn quickly, stay flexible when guidelines change, and value autonomy. If you do your best work with regular critique and a team around you, expect an adjustment.
Browse everything that's open: all micro1 roles · all Mercor roles. More platforms worth joining: AfterQuery · Ethos · Handshake
What gets you accepted
What UX and product designers need to get accepted
Several listings say outright that shipped work matters more than credentials, and DataAnnotation doesn't require a degree at all.
Almost every task ends in a written justification. Clear, specific critique matters more than polished visuals.
WCAG contrast, focus states, and usability heuristics come up again and again in grading tasks and audits.
The $80+/hr roles often ask for HTML, CSS, and React, or tools like Cursor, v0, Lovable, and Framer.
Some listings say this directly. Your design judgment is what they're paying for.
FAQ
Is AI training work for designers legitimate?
Yes. Platforms like Mercor, micro1, and DataAnnotation pay contractors to evaluate model output for AI labs. They screen applicants carefully, and some designers wait a while between projects. The legitimate platforms never charge you to apply.
Do I need experience with AI or machine learning?
No. The listings are paying for design judgment. Being able to explain clearly why a layout works or fails matters far more than any technical AI background.
What are the AI interviews like?
Short interviews run by an AI interviewer instead of a person. Most people find them awkward the first time. Treat your first one as practice, and focus on explaining your design reasoning out loud, clearly and specifically.
Are these jobs remote?
Yes. Every role we list is fully remote, work-from-home contract work, and most let you set your own hours. Some are limited to certain countries, such as US-only roles or projects open to the Americas, Europe, the UK, and Oceania. The job page always lists eligible locations.
What is the day-to-day like?
Mostly async and self-directed. You pick up tasks when they’re available and work at your own pace, with little feedback and almost no meetings. You’ll often be one of 50 to 10,000 people working on the same task, so it suits quick learners who like working independently.
Is it full-time?
Mostly not. Most roles are part-time contracts you schedule yourself, and you can work as much or as little as your availability allows, so many people work these projects on top of a full-time job. I’ve seen some specific projects enforce a 15-hour-a-week minimum during busy times, but that’s more the exception than the norm.
How long are contracts?
Most contracts are ongoing and don’t have an end date, because Mercor itself doesn’t know how long the client will want to run the project. Some last a few weeks and some run a year or more. Unfortunately, the nature of this work is unpredictable.
Is the work fun?
Some of it is fun and entertaining, like playing AI-generated games or comparing different websites. Some of it can be tedious and demanding because it’s so focused. For me, it’s cool to be on the forefront: knowing how these models work, seeing the content they create, and having a hand in shaping it.
Who are the clients?
Most of the time you won’t know who the client is until you sign an NDA and onboard onto the project. It sounds sketchy, but it’s just the nature of this industry. Many clients are frontier AI labs. You probably won’t have any direct contact with the client unless you become a reviewer or manager.
What are some of the big, reputable companies hiring for AI training?
These are the best-known platforms, ranked by how much outside funding they’ve raised, from most to least. Funding isn’t the same as size: the last two raised little or nothing because they were profitable early.
- Mercor: About $490M raised. Valued at $10B in October 2025, and reportedly in talks to raise at around $20B in 2026.
- Handshake: About $435M raised. Best known as the college career network, it now runs a large AI training program.
- micro1: About $142M raised, including a $100M+ round at a $4B valuation in September 2026.
- Mindrift: Run by Toloka, which raised $72M in 2025 in a round led by Jeff Bezos’s Bezos Expeditions.
- AfterQuery: A $30M Series A in April 2026, then a reported Series B at a $3.2B valuation (amount not disclosed).
- Ethos: $22.75M from Andreessen Horowitz in May 2026.
- Toptal: One seed round and no venture funding since. It has grown into a large, profitable business on its own.
- DataAnnotation: Reportedly run by Surge AI, which has never confirmed outside funding. It was self-funded by its founder and is reportedly valued at $15B or more.
Funding figures are from public reports as of October 2026, and some are reported rather than officially announced.
How much can a designer make training AI?
Most design roles pay $35 to $110 an hour, with labeling work starting around $15 and senior roles reaching $350. See the pay breakdown for details.
How do I get paid?
Usually weekly, as an independent contractor. Mercor pays through Stripe or Wise, micro1 pays into an account wallet you can withdraw from, and DataAnnotation pays weekly. In the US you'll handle your own taxes as a 1099 contractor. Also, some projects pay per task and some pay hourly. It depends on the project.
Don't miss the next one
Good roles fill in days
We check the platforms every day and send you only the design roles worth applying to.