RLHF & Human Preference Data
Produce preference rankings, reward-model feedback, and calibrated human judgment for post-training pipelines.
AuraOne / Help Center / AI trainer
Become an AI trainer.
An AI trainer is a subject-matter expert who grades, ranks and corrects model outputs so AI teams can train and test their models. AuraOne's jobs board lists contract roles for specialists in medicine, finance, law, engineering, languages and more.
This page explains the work, the domains, how screening and pay work, and how to apply. The terms for each role are on its listing.
The work
What an AI trainer does
Roles differ by track. Each track below has its own listings on the jobs board.
See these roles
AI Safety & Red Teaming
Push on refusal boundaries and dual-use risk before a model ships.
Frontier Model Evaluation
Review frontier model outputs. Judge benchmark failures and calibrate other evaluators.
Coding, SWE & Agent Evaluation
Judge generated code and software engineering agents. Read their debugging traces.
Medical & Clinical AI
Review clinical and biomedical AI outputs. Your safety judgment is the point.
Legal & Policy AI Evaluation
Judge legal, policy, and compliance outputs for precision and risk.
Finance, Accounting & Economic AI
Review financial reasoning and accounting treatment. Audit logic and risk work too.
Language, Localization & Culture
Review translation and localization work. Dialect and cultural context matter here.
Expert domains
Which expert domains AuraOne asks about
The specialist application asks you to choose the focus areas you work in, up to eight. These are the areas it offers. Your primary domain is free text, so you can name a field the list does not show.
Autonomy and Robotics
Perception, planning, and safety evidence.
Healthcare and Life Sciences
Clinical NLP, imaging, and specialist review.
Financial Systems
Risk, controls, and decision review.
Governance and Safety
Policy, red teaming, and audit readiness.
Content and Trust
Annotation, moderation, and policy operations.
Applied ML Operations
Data pipelines, evaluation harnesses, and regressions.
Hardware in the Loop
Simulation, edge deployments, and test operations.
AI Evaluation Science
Benchmarks, statistical validation, and scoring.
How screening works
Interview stages, the process owner, and how your data is used are stated with each opportunity.
01
Apply
4 short steps: fit, about you, experience, interests. You can apply to a listed role or create a specialist profile without choosing one.
Application received
02
Review
Our team reviews your experience, work samples, availability and specialist interests. A human reviews fit before any placement.
Team review
03
Match
AuraOne contacts you when there is a relevant opportunity, now or later.
Opportunity match
04
Onboard and qualify
If you are selected, you confirm the role, terms, required checks, access, and first assignment. On expert programs, work is assigned only after a passed test.
Onboarding and assignment
Pay and taxes
How pay and taxes work
AuraOne does not publish one rate. Pay depends on the role.
Pay is on the listing
A job listing shows its pay when the role states it. Where it does not, the listing says the hourly rate is confirmed after the interview process.
Terms come in writing
Specialists and contractors receive the rate, acceptance criteria, payment method, and payment schedule in writing before work begins.
Contract work
Roles on the board are contract roles. Placement depends on current program demand and reviewer confirmation.
Taxes in Aura Capture
AuraOne's tax article covers earnings from the Aura Capture app. If a contributor earns $600 or more in a US tax year, AuraOne files a Form 1099-NEC and sends a copy. A W-9 states the contributor's tax classification. AuraOne does not withhold taxes from payouts, and filing is the contributor's responsibility. For a contract role on the jobs board, the written terms you receive before work begins set out the pay arrangements. Read the tax forms article.
Payouts in Aura Capture
Contributors who record with the Aura Capture app earn on an earnings ledger. Payouts begin when the program turns them on. Once payouts start, each one goes to the contributor's bank through Stripe Connect after a finance admin releases it. Read the payout setup article.
How to apply
How to apply for AI training jobs
Pick a role, or create a specialist profile without choosing one. Neither action submits you to an unselected role.
What the form asks
Your name, email, location, primary domain, focus areas, experience level, years of experience, availability, and a note on your experience and motivation. You can attach a resume and add links.
Experience levels
- Emerging, 1-3 years. You have supported reviewed programs and want deeper engagements.
- Seasoned, 4-7 years. You regularly deliver models or workflows with measurable outcomes.
- Expert, 8+ years. You lead cross-functional launches and own quality for complex domains.
- Principal, Program lead. You design governance frameworks, chair reviews, or run calibration.
Start here
Before you apply
Identity checks, your data, and how to check a listing
Plain answers to the questions people ask before they share their details.
Identity check in Aura Capture
Contributors in the Aura Capture app verify identity with Stripe Identity. Stripe stores the photo of your ID and your selfie. AuraOne receives the verification status and, if a check fails, the reason Stripe gives. AuraOne never receives the images. Read the identity article.
Your application data
The privacy policy explains how applicant and contact information is handled. To delete an account and its data, see account and data deletion.
Check a listing
Start from the AuraOne jobs board on auraone.ai. A listing's role details show its track, work model, pay and employer. The company facts page lists AuraOne's legal name, address and official profiles. For anything else, use support.
Questions
Questions about AI training jobs
What is an AI trainer?
An AI trainer is a subject-matter expert who grades, ranks and corrects model outputs so AI teams can train and test their models. On AuraOne's jobs board the work spans AI evaluation, research and expert review in areas such as medicine, finance, law, engineering and languages.
How much do AI trainers get paid?
AuraOne does not publish one rate. A job listing shows its pay when the role states it. Where it does not, the listing says the rate is confirmed after the interview process. The rate, acceptance criteria, payment method and payment schedule come in writing before work begins.
Do I need a degree or a certification?
Requirements are listed with each role. The application asks for your primary domain, focus areas, experience level, years of experience and availability, and it accepts a resume and links.
How does AuraOne screen AI trainers?
Our team reviews your experience and fit, and a human reviews fit before any placement. If you are selected, you confirm the role, terms, required checks, access and first assignment. On expert programs, work is assigned only after a passed test.
How are Aura Capture contributors taxed?
AuraOne's tax article covers earnings from the Aura Capture app. If a contributor earns $600 or more in a US tax year, AuraOne files a Form 1099-NEC and sends a copy. A W-9 states the contributor's tax classification. AuraOne does not withhold taxes from payouts, and filing is the contributor's responsibility. For a contract role on the jobs board, the written terms you receive before work begins set out the pay arrangements. This is general information, not tax advice.
Is the work remote?
The specialist network offers remote AI evaluation and domain expert opportunities. Each listing shows its work model, so check the listing before you apply.
Guides
Related reading
How to Hire Domain Specialists for AI Data WorkA practical workflow for qualifying, calibrating and reviewing domain specialists for AI data, with evidence checks for buyers and prospective reviewers.Human Data Reviewers: From Resumes to Task-Specific ReputationAssess expert reviewers through calibration, adjudicated disagreements and task-specific quality records rather than resumes or a universal score.