Guide · 5 min read
Types of annotation work — requirements and onboarding
"Data annotation" covers a much wider range of work than most people expect. Some of it is simple labeling you can start this week; some of it pays specialists hundreds of dollars an hour. Below is every main type of work the platforms on this site offer, what each one asks of you, and the exact onboarding steps from application to first payout.
The types of work you can get
Text & chat evaluation (RLHF)
$15–$25/hrCompare two AI answers, rate them, and write a short explanation of which is better and why.
Requirements: Fluent written English (or a target language), careful reading. No degree needed.
Image & video annotation
$20–$50/hrDraw boxes or masks around objects, tag scenes, check whether a model's description of a clip is correct.
Requirements: Attention to detail, a decent screen, steady internet. Training provided.
Audio & language work
$20–$40/hrRecord prompts, transcribe, or review transcripts and pronunciation in your native language.
Requirements: Native-level speaker of the requested language, quiet room, phone or headset mic.
Document & business tasks
$30–$60/hrBuild or check Excel, Word, PowerPoint and PDF outputs an AI produced, or annotate business documents.
Requirements: Real office-tool experience; accounting, HR or analyst background helps.
Code & engineering
$50–$280/hrWrite tasks in real repositories, debug AI-generated code, judge architecture and system design answers.
Requirements: Working software engineering experience; senior tracks ask for years of production work.
Expert domain review
$70–$150/hrCheck AI output in your field — medicine, law, finance, physics, biology, mathematics — for factual correctness.
Requirements: Degree (often graduate level) or licensed professional experience in the field.
Live openings for each of these sit on the jobs board, or jump straight to data annotation jobs, AI training jobs and RLHF jobs.
How onboarding actually works
1. Application
You submit a profile: résumé, languages, fields of expertise, hours per week. General applications (micro1, Mercor, Ethos) mean one form for all projects instead of applying role by role.
2. Screening or AI interview
Most platforms run a short screening — an AI-led voice or chat interview, a writing sample, or a skills quiz. Usually 20–60 minutes, done once.
3. Assessment task
You complete a paid or unpaid sample task under the project's real guidelines. This is where most people fail: read the guideline document fully, take the time allowed, and never let an AI write your answers.
4. Identity & payment setup
ID verification, then Stripe, Wise or PayPal details. Platforms ask for tax info (W-9 in the US, a self-employment declaration in the EU). No legitimate platform asks you to pay a fee.
5. Matching to a project
You are placed in a queue for projects that fit your profile. This can take days to a few weeks, and the first offer is often smaller than your maximum rate.
6. Working and getting paid
You log hours or tasks in a dashboard. Most platforms pay weekly; volume comes in waves, which is why experienced people keep two or three platforms active.
Step 3 is where nearly everyone gets stuck — our assessment tips cover it in detail.
What you need before you apply
- A laptop or desktop and stable internet — phone-only work is rarely possible.
- An up-to-date résumé, including any degree or professional licence.
- A payout account (Stripe, Wise or PayPal) and your tax details.
- A quiet space and a usable microphone if you want audio or language projects.
- Realistic availability — most projects ask for at least 5–10 hours a week.
Common questions
- What are the requirements for annotation work?
- For generalist text, image and audio work: strong written English or a native language, careful reading, a laptop and stable internet. Specialist tracks paying $70/hr and up require a degree or professional experience in the field.
- How long does onboarding take?
- The application and screening usually take under an hour. Getting matched to a paid project takes anywhere from a few days to a few weeks, depending on which projects are open.
- Do I need experience to start annotation work?
- No. Generalist evaluation and labeling roles are designed for newcomers and provide guidelines and training. Experience only matters for expert domain and engineering tracks.
- Do annotation platforms charge a fee to join?
- Never. Legitimate platforms pay you; any site asking for a deposit, training fee or equipment payment is a scam.
- How do annotation platforms pay?
- Weekly or bi-weekly through Stripe, Wise or PayPal, based on logged hours or completed tasks. You work as an independent contractor, not an employee.
Start with one application
The fastest route in is a general application: one profile, one screening, and the platform matches you to every project that fits — instead of applying to each role separately.
Thousands of open slots, AI-led interview, $40–$112/hr
One application, generalist to expert projects, $15–$150/hr
Apply once, get matched to fitting projects
These are referral links — they cost you nothing and help keep this board free. Not sure about a platform? Read is micro1 legit, is Mercor legit or is Ethos legit first.