Task-based model review with the lowest entry bar in this feed, gated by one unpaid assessment.
Work type
Task queue, remote
Screening
One-shot unpaid Starter Assessment
Payouts
Weekly, PayPal
Entry bar
Lowest in this feed
19 open DataAnnotation listings in the live feed, refreshed daily. Apply links go to DataAnnotation directly; where a referral programme exists, Skillora may earn a fee at no cost to you, and it never affects what is shown. Official site: dataannotation.tech
Filtering by rate hides per-task listings, which have no hourly equivalent.
Turkish Translator Overview Automated metrics can't judge whether Turkish text actually sounds right: tone, idiom, cultural nuance, and the thousand small choices a native speaker…
Spanish Translator Overview Automated metrics can't judge whether Spanish text actually sounds right: tone, idiom, cultural nuance, and the thousand small choices a native speaker…
Software Engineer Overview The newest models can vibe-code a whole feature in minutes. Someone has to check whether that code actually works, and that someone is you. You'll run ag…
Quantitative Analyst Overview Models handle textbook finance fine and fall apart on the real thing: mispriced risk, subtly wrong derivations, and models that ignore market microstr…
Portuguese Translator Overview Automated metrics can't judge whether Portuguese text actually sounds right: tone, idiom, cultural nuance, and the thousand small choices a native sp…
Physicist Overview Models can recite physics and still get it wrong: dimensional errors, misapplied approximations, and derivations that skip the step where the physics actually ha…
Physician Overview Medical questions are where model mistakes matter most: outdated guidance, missed contraindications, and confident answers that no clinician would sign off on. A…
Mathematician Overview Frontier models still fail at rigorous multi-step mathematics: invalid derivations, unjustified leaps, and confidently wrong final answers. Automated checks…
Machine Learning Engineer Overview Models are surprisingly bad at reasoning about themselves: training dynamics, evaluation design, data pipelines, and deployment trade-offs. Plaus…
Indonesian Translator Overview Automated metrics can't judge whether Indonesian text actually sounds right: tone, idiom, cultural nuance, and the thousand small choices a native sp…
Hindi Translator Overview Automated metrics can't judge whether Hindi text actually sounds right: tone, idiom, cultural nuance, and the thousand small choices a native speaker make…
Healthcare Operations Specialist Overview Models struggle with the practical side of healthcare: billing codes, insurance workflows, scheduling, and the administrative details that…
Generalist Overview Most of what models get wrong isn't specialist knowledge. It's careful reading, clear writing, and common-sense judgment, and that's exactly what this role supp…
Data Scientist Overview The newest models can spin up a full analysis in minutes — load the data, pick a method, produce charts and a confident conclusion. Someone has to check whe…
Chemist Overview Models routinely propose impossible mechanisms, unsafe procedures, and syntheses that no bench chemist would attempt. As a Chemist you'll evaluate AI chemical reas…
Biologist Overview Models blur the line between what's established, what's contested, and what's simply made up in the life sciences, and they cite papers that don't exist. As a Bi…
Attorney Overview Models produce fluent legal analysis that is often subtly wrong: invented citations, misapplied precedent, and jurisdictional confusion. A plausible wrong answer…
Arabic Translator Overview Automated metrics can't judge whether Arabic text actually sounds right: tone, idiom, cultural nuance, and the thousand small choices a native speaker ma…
Accountant Overview Models produce confident accounting answers that fail on the details: misapplied standards, broken reconciliations, and journal entries that don't balance. As a…
DataAnnotation.tech is the entry point most people find first. You create an account, pass one unpaid Starter Assessment, and then pick tasks from a queue: writing and ranking model responses, fact-checking, and a separately-gated coding track that pays roughly double the base rate.
It is task-based rather than role-based. There is no project interview and no fixed schedule; you work the queue when work is there and stop when it is not. The company keeps its client list private and communicates sparsely, which is the trade for the lowest barrier to entry in this feed.
Payment has a long, clean track record: weekly, via PayPal, from the first task you complete.
$25–150
per hour, published range
$100/hr
median listing
19
live listings publishing a rate
Computed from DataAnnotation’s own published rates on the live feed, not from survey estimates.
Published rates start around $20 an hour for general work; the coding queue posts $25 to $40, and specialist projects above that. Rates on the live listings below are shown exactly as DataAnnotation publishes them, and pay lands weekly via PayPal.
Generalists with sharp writing, and programmers. The Starter Assessment cares whether you follow instructions exactly and write careful, natural English; the coding assessment unlocks the higher-paid track, and Python or JavaScript covers most of it.
No degree requirement, no professional-credential screen. That makes DataAnnotation the right first platform for strong writers without a licensed specialty, and a common second income for engineers who clear the coding tier.
The Starter Assessment is unpaid, takes about an hour, and you get one attempt. No retakes, no feedback, and usually no rejection email: silence for two or three weeks means no.
That one-shot structure should change how you take it. Do it fresh, read every instruction twice, and write complete answers in the plain, precise register the tasks themselves demand. People fail it by rushing, by padding answers, and by ignoring formatting rules that were stated once. Treat it like the paid work it gates.
Skillora runs the same format: a domain interview with an AI interviewer, followed by feedback on how you explained your reasoning. Take it as many times as you want before the one that counts.
Start a free practice interviewYes. It has years of contributor payment history and pays weekly through PayPal. The frustration people report is not about money: it is the silent rejection after the Starter Assessment and the opaque way projects appear and disappear. If you are accepted, the platform does what it says. If you are not, you will simply never hear back.
The base tier starts around $20 an hour for writing, ranking and fact-checking tasks. The coding queue posts roughly $25 to $40 an hour, and occasional specialist projects run higher. The live cards above show current published rates. There is no negotiation; the rate is attached to the task.
No. One attempt per person is the standing policy, and duplicate accounts get banned. That is why the preparation advice everywhere, including here, is the same: take it rested, read the instructions like a contract, and write every answer as if a grader is scoring instruction-following first and prose second. They are.
Accepted contributors typically see projects appear within days to a couple of weeks. There is no rejection notice: an empty dashboard after two to three weeks is the answer. At that point your time is better spent on the other platforms in this feed than on waiting.
A separate gate after the starter, testing whether you can read, write and debug code well enough to judge model output: mostly Python and JavaScript, with an emphasis on explaining what is wrong and why. Passing it roughly doubles your rate, so engineers should attempt it immediately rather than settling into the generalist queue.
Steadier than Outlier for generalists, but still demand-driven. Projects surge and pause without notice, and the queue can thin out for days. Contributors who treat it as one feed among several, and who hold both the writing and coding qualifications, see the most consistent weeks.
