Pick the work that fits. One side never touches AI.
Take a short assessment and get an answer in 72 hours. Every applicant hears back. Web Work starts at $65/hr and AI Work at $85/hr.
Placing contributors since 2024
$6.2M+
Generated for contributors
1,000+
Active jobs
$0
Fees. Applying and working cost nothing
72h
Decision deadline, every applicant
Two tracks
Pick the one that matches how you want to work.
AI Work · selective
AI Work
You read model output and judge it. The skill being bought is skepticism.
| General evaluation | $85/hr |
| Rubric & comparison work | $105/hr |
| Code review | $120/hr |
| Specialist domains | $135/hr |
6 questions · 20 min · decision in 72h
What AI Work looks like →Web Work · open
Web Work
Zero AI contact at any point. You verify, check, transcribe, and translate.
| Listing & catalogue QA | $65/hr |
| Content moderation | $70/hr |
| Research & verification | $75/hr |
| Transcription · translation | $80/hr |
4 questions · 12 min · decision in 72h
What Web Work looks like →Not sure which fits you? Pick “not sure” on the form and the assessment routes you to the track your answers fit best.
The work itself
Six real tasks.
Representative of the work in each queue. Times are estimates, not measured averages.
~8 min
AI Work
Read two answers to the same question about a Python error message. One is right. Explain in two sentences what the other gets wrong.
~12 min
AI Work
Score a model's summary of a 900-word news article against a five-point rubric. Mark each point pass or fail with one sentence of justification.
~15 min
AI Work
Review a 40-line function that merges date ranges. Its tests pass. Find the input the tests miss and describe the failure in two sentences.
~6 min
Web Work
Confirm whether a dental clinic in Porto is still operating. Use the official business register, not a maps listing.
~45 min
Web Work
Check 30 product listings against a 14-rule style guide. Flag each violation with the rule number it breaks.
~50 min
Web Work
Transcribe a 12-minute interview with two speakers. Label speakers, timestamp every 30 seconds, mark unclear audio.
Rates
What each task pays.
Vague pay ranges read as a scam in this category, because they usually are one. Here is the whole table.
Payout schedule and probation pay →AI Work
Selective. 6-question assessment, about 20 minutes.
| General evaluation | $85/hr |
| Rubric & comparison work | $105/hr |
| Code review (Python, JS, TS) | $120/hr |
| Specialist (law, medicine, finance, advanced STEM) | $135/hr |
Web Work
Open. 4-question assessment, about 12 minutes.
| Listing & catalogue QA | $65/hr |
| Content moderation | $70/hr |
| Research & verification | $75/hr |
| Transcription & translation | $80/hr |
Paid twice a week
On the two payout days you pick. Hours count on task time, so a hard task never cuts your rate.
Probation pays full rate
All 50 tasks of a Web Work probation batch pay the same as the open queue.
Rates posted per queue
Before you take a task. A rate change never applies to work already done.
From people doing the work
What it is actually like.
I've been working on projects aligned with my professional background that challenge and motivate me to create tasks and rubrics to improve use of AI.
I joined EYAS and have been active there in the last month. It is a great place for a side job, and it is possible to earn upward of $1000 per week if you put in the work.
It is really enjoyable to work in Eyas. I can work at anytime and place and the pay rate is awesome. I really recommend it!
It has been fulfilling to contribute to the ongoing digital revolution, particularly in the field of AI model development.
Never thought I could be making this much with my little to no-experience in Web in general, can make up to 5k per month. I'm really glad I applied.
Earnings depend on track, hours, and task type. Rates are posted per queue; individual results vary.
An eyas is a young hawk still in training. Everyone here starts as one.
How it works
Five steps.
The full process →Apply
3 min
Email, country, languages.
Assess
20 or 12 min
One question per screen. No timer.
Decision
72h
Accepted, routed, or declined.
First work
7 days
Live queues, or a paid probation batch.
Ongoing
2× weekly
Take a task, finish it, take another.
Careful judgment is a skill you can test for directly. That is the whole idea.
Trust
Graders never see your name.
Work platforms run on silence: you apply, then nothing. We commit to publishing our pipeline numbers, and the database is built so identity and answers cannot meet.
Who we are and who pays →Anonymous grading
Graders score your assessment under an anonymous ID. They never see your name or email.
Two demographic fields
Country and languages, because projects require them. We never collect gender, race, or ethnicity.
72 hours, in writing
Every applicant gets a decision email inside 72 hours, accepted or not.
Disclosed telemetry
During the assessment: per-question timing, typing and paste activity, window focus. Nothing outside it.
FAQ
Asked most often.
All 12 questions →- Do I need a degree?
- No. There is no resume screen. The assessment takes 12 or 20 minutes and is the entire application.
- Is Web Work really AI-free?
- Yes. No AI content in the application, the assessment, or the work. Zero contact at any point.
- How much will I earn?
- AI Work pays $85 to $135/hr by task type. Web Work pays $65 to $80/hr. Every queue posts its rate before you take a task.
- When do I get paid?
- Twice a week, on the two payout days you pick. Probation tasks pay the full queue rate.
- What if I'm declined?
- Your email includes a retry date 60 days out. Questions rotate, so the retake is a different set.
- What happens to my data?
- Name and email in one table, answers in another under an anonymous ID. Deletion on request within 30 days.