Platforms Tracked7 + direct desk
Open Referral Rails3 · Class A
Expert Roles Up To$200/hr
Top Referral Payout$3,000
Programs VerifiedJuly 2026
July 2026 · Programs Verified

Train the
models.
Get paid
for it.

Seven AI-training platforms, one desk. Real pay ranges, real entry bars, a 30-second matcher — and every referral link on this page labeled by class, so you always know which ones pay us.

Affiliate disclosure · some links pay this site a referral fee when a candidate is placed, at no cost to the candidate. These are independent-contractor marketplaces, not employment offers; we are not a recruiter or employment agency.

Platform Matcher

Three questions. One recommendation, with the reasoning shown.

1 · Your background
2 · Where you work from
3 · What matters most
Your match

Matches follow a fixed, published decision tree — they never depend on whether a link pays us. Each platform's referral class is disclosed on its card.


The Desk · July 2026

Pick your platform.

Every card shows real pay ranges, the entry bar, and — unusually for this genre — exactly how this page is or isn't compensated. The class system below is the house rule: recommendations never follow the money, and the money is always labeled.

Class A
Open referral rails

Programs anyone may join as a referrer. Links become referral-tracked once configured; the platform pays this site on qualified placement, at no cost to you.

Class B
Contributor-gated

Referrals require an active contributor account, and program terms constrain public posting. Links here stay plain and unmonetized until each program's terms are cleared.

Class C
Direct application

No referral program exists. Listed on equal footing anyway — a desk that only showed platforms that pay it wouldn't be worth reading.

Program verified
Highest-Volume Desk
micro1Class A
$15–$150+
per hour by role — engineering listings at the top of the range
WorldwideHundreds of live rolesAll backgrounds

The largest listing volume in AI training. A 15–25 minute AI-led interview (their recruiter, Zara) gates entry; clear it and you're matched across generalist evaluation, domain-specialist, and $100+/hr engineering roles. The realistic first stop for most people.

Open referral program — micro1 pays this desk up to $3,000 when a referred candidate is placed and completes 10 hours. Your pay is unaffected.
Browse micro1 roles

Entry: AI interview + possible role-specific assessment. Candidate pay is listed per role. Certification does not guarantee placement. Independent-contractor engagement; worldwide, role permitting.

Program verified
MercorClass A
$100–$200
per hour on expert listings — the highest top end in the space
Domain expertsWorldwideCurated roles

A smaller, curated pool of frontier-lab projects that pays a premium for credentialed expertise — physicians, attorneys, PhDs, senior engineers. If you have a specialty, run Mercor in parallel with micro1 for the higher ceiling.

Open referral program — Mercor pays this desk a share of a referred expert's earnings, capped by their first offer. Your pay is unaffected.
Browse Mercor roles

Entry: resume + AI interview; selective matching to lab projects. Generalist roles pay below the expert range shown. Independent-contractor engagement.

Program verified
TuringClass A
Long-term
U.S.-client engineering contracts — months, not micro-tasks
EngineersWorldwide talentFull-time-style

The structural opposite of gig tasking: vetted engineers matched into ongoing contracts with U.S. companies, increasingly on AI and LLM work. Slower to land, far stickier once landed. Compensation varies widely by role and region.

Partner referral program — Turing pays this desk per developer who completes a first month with a client. Your pay is unaffected.
Browse Turing roles

Entry: skills vetting + matching cycle; timelines depend on client demand. Independent-contractor engagement with Turing's clients.

Program verified
Handshake AIClass B
Up to $125
per hour, project-based — the Fellowship for students, grads & pros
U.S. onlyWork auth requiredWeekly pay

Flexible LLM-training projects for U.S.-based bachelor's, master's, and doctoral students, candidates, and graduates — including F-1 students on CPT or OPT (STEM OPT excluded). ID verification at signup; matching follows your background.

Contributor-gated referral program — this link is plain and unmonetized until the desk holds a cleared contributor account.
Open the Fellowship

Entry: Handshake account + profile + government-ID verification. U.S. work authorization required; confirm visa conditions with your program sponsor before applying.

Program verified
OutlierClass B
$10–$60
per hour typical — specialist tracks reach higher
WorldwideScale AIWeekly Tuesday pay

Scale AI's contributor platform: prompt writing, response ranking, and domain tasks across math, code, writing, and specialist fields. Task availability fluctuates by project cycle; payment runs weekly via PayPal, Airtm, or ACH.

Referrals are dashboard-gated and framed as personal vouching — this link stays plain and unmonetized pending terms clearance.
Open Outlier

Entry: associate degree minimum; selective screening, higher bars on specialist projects. Independent-contractor engagement.

Program verified
AlignerrClass B
High-rate
selective expert projects — rates up to ~$150/hr reported
ExpertsSelectiveLabelbox

Labelbox's expert network for model training and evaluation — audio, code, and domain-specialist projects. A skills assessment at onboarding determines which project tiers you can claim; acceptance is deliberately narrow.

In-platform Refer & Earn — gated to contributors; this link stays plain and unmonetized pending terms clearance.
Open Alignerr

Entry: registration + skills assessment + profile review. Third-party rate reports; verify current figures in-platform. Independent-contractor engagement.

Program verified
DataAnnotationClass B
$20–$30+
per hour commonly reported — varies by project and qualification
Entry-friendlyWriting & coding tasksFlexible

Chat-based model feedback and evaluation tasks with one of the lowest entry bars on this desk. A starter assessment gates access; project flow arrives in waves, so treat it as supplemental rather than steady income.

Worker-gated referral codes issued in limited waves — this link stays plain and unmonetized pending terms clearance.
Open DataAnnotation

Entry: signup + qualifying assessment; acceptance timelines vary. Independent-contractor engagement.

Class C Direct-application desk

No referral programs exist for these — every link above is plain. They're listed because a credible desk covers the field, not just the platforms that pay it. Surge is invite-selective; Prolific pays for research participation rather than training work.


AI Career Launchpad · Talent Pool

The AI build-out is hiring. Get in the pool.

The platforms above are the front door. The Launchpad is the direct channel: one verified profile that we present to AI training companies, enterprises, and recruiting partners negotiating contracts right now — with your consent versioned, hashed, and revocable.

One profile.
Every rising tide.

Frontier labs are paying premium rates for human expertise, and demand is moving faster than any single platform's intake. The Launchpad puts you in front of the whole current — without re-screening for every desk.

Matched, not blasted. Your profile is presented to partners whose live contracts fit your background — never dumped into a bulk list.
Free for candidates, always. Partners compensate this desk; you are never charged, and your pay is never reduced by our involvement.
Consent with receipts. The consent text below is versioned and fingerprinted; your submission returns a receipt hash, and withdrawal or deletion is honored on request.
Minimum viable data. We collect what matching requires — nothing else. No government IDs, no date of birth, no documents beyond your resume.

Joining the pool is not a job offer and does not guarantee introductions, engagements, or income. Opportunity flow follows partner demand. 18+ only.

Work eligibility (for matching only)
Resume / CV *
You're in the pool.
Keep this receipt — it proves exactly what you consented to and when. Reply to any introduction email, or write privacy@predictionmarkets.xyz, to update, withdraw, or delete at any time.
Privacy Notice — AI Career Launchpad
Who we are. The Launchpad is operated by PredictionMarkets.xyz ("we"). Contact: privacy@predictionmarkets.xyz.

What we collect. Only what you submit above: name, contact details, country, professional links, background, specialties, languages, availability, target rate, work-eligibility regions, and your resume. We do not collect government IDs, dates of birth, or any special-category data, and we ask you not to include such data in your resume.

Why. To evaluate your profile and introduce you to AI training companies, enterprises, and recruiting partners with relevant engagements — the service you are requesting — and, if you separately opt in, to send occasional opportunity emails. Partners compensate us for introductions; candidates are never charged.

Sharing. Your profile and resume are shared only with prospective engaging partners in those categories, for placement purposes, under confidentiality terms. We do not sell personal information to data brokers and do not use it for advertising. California residents: introductions to partners may constitute "sharing/selling" as those terms are defined by the CPRA — you may opt out or delete at any time via the contact above, and we honor those requests without discrimination.

Retention. Profiles are kept for 24 months from your last update, then deleted unless you refresh them. You can request access, correction, portability, or deletion at any time; EU/UK residents may also withdraw consent and lodge a complaint with their supervisory authority.

Consent integrity. The exact consent text you accept is versioned and fingerprinted (SHA-256, shown beneath the checkboxes); your submission receipt binds your consent to that version. 18+ only.

Contractor Toolkit · Tax & Reporting
Every platform above pays you as a 1099 contractor.

No withholding, quarterly estimates on you, and income arriving from three platforms at once by Q2. Koinly consolidates payout records alongside any crypto proceeds and produces filing-ready reports — the same toolkit rail as our prediction-markets desk.

KoinlyRecurring
20% off
Your first plan — affiliate-linked; disclosure applies
Get Koinly

The Worker Journey

How getting placed works.

Every platform on this desk follows the same four-stage pattern. Where they differ is the entry bar and the pace of project flow.

01
Pick your platform

Use the matcher or the comparison table. Active contributors typically run two or three platforms in parallel — apply to your match first, add the runner-up once accepted.

02
Clear the screen

micro1 and Mercor run AI-led interviews on your actual resume — concrete, specific answers pass. Others use skills assessments. Handshake adds government-ID verification.

03
Qualify for projects

Matching follows your verified background, and most platforms add short project-specific onboarding or guideline exams. Placement is never guaranteed — flow follows lab demand.

04
Work and get paid

Hourly or per-task, as an independent contractor. Weekly cycles are the norm — Outlier pays Tuesdays; Handshake pays weekly. No withholding: set aside for quarterly taxes.


Side-by-Side

Compare the desk.

Pay ranges reflect verified figures and commonly reported rates as of July 2026; each platform's live listings control. Click a column header to sort.

Platform Focus Pay Range Geography Entry Bar Referral Class
micro1 Everything — generalist to engineering $15–$150+/hr Worldwide AI interview (15–25 min) A · Open
Mercor Credentialed domain experts $100–$200/hr (expert) Worldwide Resume + AI interview, selective A · Open
Turing Long-term engineering contracts Varies by contract Worldwide talent, U.S. clients Skills vetting + matching cycle A · Open
Handshake AI Students, grads & professionals Up to $125/hr U.S. + work auth only Profile + government-ID check B · Gated
Outlier Flexible tasking, all domains $10–$60/hr typical Worldwide Assoc. degree min, screening B · Gated
Alignerr Selective expert projects Up to ~$150/hr reported Worldwide Skills assessment, narrow intake B · Gated
DataAnnotation Entry-level writing & coding $20–$30+/hr reported Worldwide, wave-based access Starter assessment B · Gated
Surge · Appen · Prolific · xAI Direct applications Varies Varies Platform-specific C · Direct

Program Economics · Full Transparency

How this desk earns — modeled openly.

The same disclosure standard we apply to prediction-market bonuses: exact program mechanics, an estimator with printed assumptions, and a hard line between rails that are live and rails that stay dark until terms clear.

Referral economics estimator

Estimates only — program terms control, every figure assumes fully qualified placements, and Class B rows are shown dark until their programs are cleared and enrolled.

Qualified placements / mo 8
Candidate profile
Program Basis Est. monthly

Assumptions: micro1 payouts are listed per role — modeled at $500 for generalist-tier and $2,000 for expert-tier listings against a published ceiling of $3,000. Mercor modeled at its hourly-contract cap (4× the first accepted offer's hourly rate); marketed range runs $250–$15,000 where deliverable-contract caps apply. Turing modeled at the $1,000 talent fee, engineering placements only, paid after a completed first month; client referrals ($2,000–$20,000) excluded. Handshake modeled at its cap (10% of rate × 40 hours) and requires an active contributor account. All programs attribute at signup via link only — no retroactive claims.

micro1 Class A

Open to any referrer — no platform account or certification required. Up to $3,000 per referral, with the exact payout listed on each job posting before you share. Triggers when the candidate is hired and completes 10 hours (or one task on per-task projects) within 3 months; paid to a withdrawable wallet; no cap on referrals. Enroll at refer.micro1.ai. A second rail pays up to $50,000 for referring companies into their Data Partnerships program.

Mercor Class A

20% of everything a referred expert earns — hourly, bonuses, one-time payments — capped at 4× the hourly rate of their first accepted offer; marketed at $250–$15,000 per referral. Enroll at work.mercor.com/refer. Terms require consent-based, non-spam outreach; a passive board where candidates click through qualifies, mass messaging does not. Work-trial and Rubric Academy contracts are ineligible, and Mercor holds clawback discretion on suspect attribution.

Turing Class A

Formal partner terms: $1,000 per talent referral, paid when the developer completes a first month with a client, plus client-referral fees of $2,000 up to $20,000 per developer started. Enrollment required, with published content standards on referral postings — comply before wiring the rail. The in-network developer tier (minimum $250 per match, no limit) is the fallback if partner enrollment lags.

Handshake AI Class B

Contributor-only: the referrer must hold an active Handshake AI account at referral and at payout. Pays 10% of the referral's hourly rate per hour worked, unlocking at 10 hours, capped at 40 hours per referral, inside a 6-month window — an effective ceiling of 4× the referral's rate. Weekly payout; anti-gaming reserve. Rail stays dark until this desk holds a cleared account.

Outlier & Alignerr Class B

Outlier: referral links live in the contributor dashboard and official guidance frames referrals as candidates you can personally vouch for — the worst structural fit for an anonymous funnel, so this rail stays dark pending written clearance. Alignerr: in-platform Refer & Earn with reported milestones around $25 at onboarding and $125 at 10 hours worked; verify current figures in-dashboard before enabling.

Attribution mechanics All programs

Every program on this desk attributes strictly at signup via the referral link — no retroactive claims, anywhere. A candidate who applied independently before clicking is unattributable. Deep-link integrity is therefore the whole game: links route directly to application endpoints, and this page never interposes redirects that could strip parameters. Candidate compensation is identical with or without a referral.


Tezos X Trust Layer

Don't trust this desk. Verify it.

The programs sheet — every pay figure, payout term, and verification date on this page — is a canonical dataset your browser fingerprints on load. Anchor it to Tezos and any post-publication edit becomes provable. Placed contractors can sign attestations with their own wallets.

Proof of Publication
Programs-sheet fingerprint

SHA-256 of the embedded programs sheet, computed live in your browser. When an anchor is declared, the fingerprint is checked against the published hash and the anchoring operation is looked up on-chain.

Computing fingerprint…
sha-256
Anchor op not yet anchored
Anchored at

Fingerprint covers the canonical #programs-sheet JSON — the same dataset that renders the cards, matcher, and verified dates above. Same anchoring flow as the prediction module. Chain reads via TzKT — Powered by TzKT API.

Placed & paid — signed by contractors

Wallet-attested reports of real placements and payouts — signed with octez.connect, attributable to a real Tezos address. Signing is free, stays off-chain, and no wallet is ever asked for funds.

No signed attestations yet.


Common Questions

FAQ

What is AI training work, exactly?
Paid, remote, project-based work that improves AI models: writing and ranking model responses, red-teaming outputs, expert review in domains like medicine, law, finance, and code, plus classic annotation and evaluation. Platforms match independent contractors to projects from frontier AI labs. Pay is hourly or per task, schedules are flexible, and the work is real knowledge work — vague answers fail the screens, concrete expertise passes them.
Is this legitimate? Does it actually pay?
The platforms on this desk are established marketplaces run by funded companies — micro1, Mercor, Turing, Handshake, Scale AI's Outlier, Labelbox's Alignerr, and DataAnnotation — paying on weekly or regular cycles through standard rails. What it is not: guaranteed income. Acceptance is selective, project flow follows lab demand, and rates vary by qualification. Treat it as flexible contract work, and never pay anyone to apply — legitimate platforms charge candidates nothing.
Who actually gets accepted?
The screens select for demonstrable expertise. AI-led interviews (micro1's Zara, Mercor's) probe the specifics on your resume — metrics, projects, real examples pass; generic answers fail. Domain credentials (a license, a PhD, senior engineering experience) unlock the highest-paying tiers. Generalists get accepted too, especially at micro1 and DataAnnotation, at lower rates. The standard play is applying to your matcher result first, then adding the runner-up once you're in.
Can I do this outside the United States?
Mostly yes. micro1, Mercor, Turing, Outlier, Alignerr, and DataAnnotation accept international contractors for many roles, with availability differing by project and region. The exception is Handshake AI, which requires being U.S.-based with valid work authorization — including F-1 students on CPT or OPT, but not STEM OPT. Always confirm eligibility on-platform before investing screening time, and visa holders should confirm conditions with their sponsor first.
How does this site make money from job listings?
Through referral programs the platforms themselves run: when a candidate clicks a referral-tracked link, gets placed, and completes qualifying work, the platform pays this site — the candidate's pay is identical either way. The class system makes this auditable: Class A links are open programs and referral-tracked once configured; Class B programs gate referrals behind contributor accounts and posting restrictions, so those links stay plain until terms clear; Class C platforms have no program and are listed anyway. This disclosure follows FTC affiliate guidelines.
How does the Platform Matcher decide?
A fixed, published decision tree — never the money. U.S. students and recent grads route to Handshake AI; international students route to micro1. Credentialed experts prioritizing rate route to Mercor. Engineers seeking long-term contracts route to Turing. Generalists, and anyone prioritizing fastest acceptance, route to micro1 on listing volume. Every result shows its reasoning and a runner-up, and the referral class of the destination is disclosed on its card.
What do the usage counts on this page represent?
When a usage count appears next to "Program verified," it is the real number of times visitors have opened that platform from this page since launch, recorded by the page's own counter. Counts start at zero, are never seeded or simulated, and stay hidden until genuine activity accumulates. If no shared counter is configured, no count is shown at all. When anchoring is enabled, a daily digest of the counts is fingerprinted to Tezos so history can't be quietly rewritten.
What is the Tezos X Trust Layer doing on a jobs page?
Two things. Proof of publication: the programs sheet — every pay figure and payout term on this page — is fingerprinted with SHA-256 in your browser on every load, and once anchored to Tezos, any post-publication edit is provable. Signed attestations: placed contractors can connect a Tezos wallet through octez.connect and sign a short report of a real placement or payout — free, off-chain, attributable to a real address, and verified server-side before publication. Device-local reports are labeled as such. This site never asks a wallet to move funds.
What is the Career Launchpad, and what happens to my data?
A voluntary talent pool. You submit one profile and resume; we present it — under confidentiality terms — to AI training companies, enterprises, and recruiting partners whose live contracts fit your background, and those partners compensate this desk for introductions. You are never charged, your pay is never reduced by our involvement, and we collect only what matching requires: no government IDs, no dates of birth, no special-category data. The exact consent text is versioned and fingerprinted, your submission returns a receipt hash, profiles are retained 24 months from last update, and we never sell to data brokers. Joining guarantees consideration, not placement.
How do I update, withdraw, or delete my Launchpad profile?
Email privacy@predictionmarkets.xyz from your signup address — or reply to any introduction — to access, correct, export, or delete your profile, or to withdraw consent going forward. Verified deletion requests are processed within 30 days. California residents may use the same channel to opt out of "sharing/selling" as the CPRA defines those terms; EU/UK residents may additionally lodge a complaint with their supervisory authority. Exercising any right never affects how you're treated.
What about taxes?
Every platform here engages you as an independent contractor — 1099 in the U.S. — meaning no withholding, self-employment tax, and quarterly estimated payments once income is meaningful. Keep payout records from day one; running two or three platforms at once multiplies the bookkeeping fast, which is why the toolkit rail exists. This is general information, not tax advice — a professional who knows your situation controls.

Affiliate Disclosure. PredictionMarkets.xyz earns referral fees when candidates placed through referral-tracked links on this page complete qualifying work, at no cost to the candidate. Monetized links are labeled by class; unmonetized platforms are listed on equal footing; matcher recommendations do not depend on compensation. Disclosure made per FTC affiliate marketing guidelines.

The opportunities listed are independent-contractor engagements on third-party marketplaces — not employment offers — and the AI Career Launchpad is a voluntary introduction service, not employment. This site is not an employer, never charges candidates, and does not guarantee acceptance, introductions, project availability, rates, or income for anyone; opportunity flow follows platform and partner demand. Each platform and partner sets its own eligibility, screening, compensation, and terms, which control. Visa holders should confirm work-authorization conditions with their sponsor before applying. Contractor income is generally subject to self-employment tax; nothing here is tax or legal advice.

Pay figures combine platform-published rates and commonly reported third-party figures, verified as of July 2026; live listings control. Usage counts reflect real visitor interactions on this page only. On-chain reads powered by TzKT API. Connecting a Tezos wallet on this page is optional, free, and never involves a transfer of funds.