OpenTrain AI

What is OpenTrain AI?

OpenTrain AI is a data labeling marketplace where AI companies hire human experts for RLHF, LLM evaluation, red teaming and annotation, and where freelancers look for AI training work. It supplies people rather than annotation software, and handles job posting, screening, contracts and payment.

The company positions itself as a transparent talent marketplace connecting AI companies with vetted trainers and data labelers, with published pricing, live-proctored vetting and one platform across 20+ annotation tools. Its Privacy Policy names the operator as OpenTrain AI, Inc., and the About page lists Weston Dotson as Founder & CEO.

Buyers choose between self-service hiring and a managed service run by a dedicated program lead; freelancers join free, and data labeling companies can list themselves to bid on client projects.

The headline network figure, 511,000+ domain experts across 40+ fields and 230+ countries, is the vendor's own count, and other official pages show different numbers (see Limitations).

Core features

Job posting and AI-matched shortlists

A client posts a job, gets a curated shortlist and brings the chosen experts into any platform the team uses. According to the vendor, its matching system draws on performance data from across 20+ AI training platforms, and every applicant goes through AI interviews and skill testing before reaching the shortlist. The job wizard asks for data type, subject matter, task type, tools, volume, location rules and languages, and these fields drive both matching and screening.

AI interview screening

• Chat-based interview: the AI interview is a chat inside the proposal modal, where an AI interviewer asks applicants questions one at a time before they can bid.
• Employer-defined fields: for recruiting-configured jobs, structured extraction can collect a LinkedIn profile, years of experience, availability or custom answers, which appear in a screening table and CSV export.
• Advisory signals: the interview can use resume and profile context to flag mismatches, but these signals help review and do not replace the hiring decision.

Work inside existing annotation tools

OpenTrain's FAQ states that it provides people, not tools. Experts work directly inside platforms such as Label Studio, Argilla, Encord, V7 Labs, CVAT or the client's own tooling, and the company says it never requires proprietary software. The integrations directory presents each listed tool as a place to hire AI trainers directly into.

Workspace and milestone payments

• Team workspace: chat, shared instructions, training modules and progress tracking sit in one place.
• Milestone payments: employers fund work in advance, trainers complete deliverables, and funds are released through Stripe once the employer approves.

Freelancer job feed and profile

For freelancers, OpenTrain pulls AI training jobs from 20+ platforms into one feed. A single profile lists skills, domains, languages and rates, and that full profile travels with each application made through OpenTrain.

Guide

Hiring through self-service

1. Post the job. Start a new job from the employer dashboard. If a brief already exists, paste it or attach a PDF and OpenTrain uses it to autofill later wizard fields.
2. Configure screening. Write the AI interview requirements and choose the expected experience level.
3. Choose a payment model. Hourly, fixed-price or per-label pricing can be used where available. Nothing is charged at publication, because funding happens when you hire and create the first milestone.
4. Hire from proposals. The first milestone is created during the hire flow, and the amount and description entered while hiring become that trainer's initial funded milestone.
5. Add and approve milestones. Each later milestone requires a description of the deliverable, an amount in USD and an optional due date. Approving a completed milestone starts the payout to the trainer.

Getting paid as an AI trainer

1. Apply from the job feed; jobs with screening open the AI interview before the bid step.
2. Connect a payout method under Settings → Payouts. Setup runs through a Stripe-hosted form that collects identity, tax and bank details, and trainers are told not to create a separate Stripe account.
3. Request payout by clicking Complete & Request Payout on a funded milestone, which then moves to Pending approval while the employer reviews the work.
4. Receive funds after approval. OpenTrain uses a daily payout schedule, and processing times vary by country and bank.

OpenTrain AI use cases

These are the workloads OpenTrain markets; results depend on the experts a client hires.

• RLHF and preference data: RLHF experts handle preference ranking, DPO pair generation and reward model training data.
• Reasoning verification: expert evaluation of chain-of-thought reasoning, process reward annotation, and verification of math, code and logic outputs.
• Red teaming and safety: adversarial prompt crafting, jailbreak discovery, bias evaluation and safety alignment testing.
• Domain expert annotation: PhDs and professionals in mathematics, medicine, law, finance and engineering supplying expert-grade training data.
• Agent training: listed solutions include function calling, computer-use agent training, agent simulations and RL environment design.
• Classic data labeling: document, image, video, text, audio, time series, point cloud, medical imaging and satellite imagery annotation.
• Pilot to production: many clients start with 5-10 experts for a pilot project and scale to hundreds for production.

Who is it for

• AI teams with annotation managers: the FAQ recommends self-service when an internal team can manage annotators directly and wants to control hiring.
• Teams without operations capacity: managed service is recommended when OpenTrain should run recruiting, vetting, onboarding, QA and team management.
• Freelancers seeking AI training jobs: one profile covers job listings from 20+ platforms.
• Data labeling companies: vendors can apply for projects posted by AI companies and draw on the talent network for recruiting.

It fits less well in a few situations:

• Minors: users must be at least 18 years old to use the Services.
• Highly sensitive datasets: the Terms say users should not upload regulated health information (including PHI) or other highly sensitive data unless explicitly authorized in writing with appropriate safeguards.

Platforms

• Web app: users sign up at app.opentrain.ai as an employer or as an AI trainer, and the account type determines the dashboard and available features.
• Android: as captured on October 5, 2026, the store listing is rated Everyone, shows 1K+ downloads and names WMD GLOBAL, LLC as developer. It supports screening that includes voice interviews, plus photo, video and audio data-collection tasks that queue offline.
• iOS: the App Store app is free, rated 4+, requires iOS 15.1 or later and lists WMD Global, LLC as seller. Both store names differ from the OpenTrain AI, Inc. named in the Privacy Policy; the official pages reviewed do not explain the relationship.
• Agent and developer access: OpenTrain says software, including fully autonomous coding agents, can hire, manage and pay AI trainers end to end. An MCP server, a CLI and a raw HTTP API talk to the same API and share the same account, tokens and permissions. For a claimed account, anything that hires someone or moves money still pauses for approval in the web app.

Pricing

Self-service clients set the rates, schedules and guidelines. Managed pricing is the team's rate plus a flat 20% management fee, with no contracts and no minimums stated, and scope can be recruiting only, full operations, QA only or anything in between.

A few billing rules sit outside the table:

• Subscriptions: agency and recruiting-configured accounts may require a subscription plan, billed separately from individual milestone charges; its price was not shown on the pages reviewed.
• Refunds: OpenTrain fees are non-refundable unless required by law or stated in a separate policy, and disputes about work quality or scope are primarily between the parties to the contract.
• Fee types: the Terms list talent-side fees among possible fee types and let fee changes apply prospectively, while the freelancer page promises "100% free for freelancers. Always."
• Conflicting fee statistic: the homepage also displays "15–20% Flat fee. Published, transparent." beside FAQ and pricing text that give the separate self-service and managed percentages shown in the table.

OpenTrain AI alternatives

The feed illustration on OpenTrain's freelancer page labels Mercor, Handshake, Micro1, DataAnnotation and Outlier as job sources, so several are both feed sources and alternatives. On OpenTrain the hiring company sets rates and runs work in its own tools; the options below present themselves as operators of their own expert work or data programs.

• Outlier: a platform operated by Scale AI that connects experts with leading AI companies to provide human feedback that improves large language models. It invites MAs, PhDs and college graduates to work as AI trainers and says no AI experience is needed.
• DataAnnotation: advertises remote contract roles matched to skills, with consistent projects, flexible hours and expert rates of up to $75–$125+/hr.
• micro1: describes itself as a data lab building infrastructure for advancing intelligence through expert human data, real-world training environments and contextual evaluations. Its Realm product offers frontier evaluations and RL environments for AI labs, work that OpenTrain also lists among its solutions.

Limitations

Marketplace role and guarantees

• No outcome guarantee: the Terms do not guarantee that a client will find suitable talent or vendors, that talent will be selected, or that any project will be completed successfully.
• Not an employer: OpenTrain does not employ talent and does not guarantee any minimum amount of work, income or engagement.
• Classification risk sits with clients: clients are solely responsible for deciding whether talent is engaged as contractors or employees and for wage, tax, benefit and insurance law.
• Vetting depends on the plan: live domain-specific testing with human proctors and a live interview applies to managed service; in self-service you get a pre-screened, AI-matched pool and run your own evaluation.

Payments and payouts

• Payout countries: trainer payouts only reach countries on a published list (last updated 2026-05-01). If a country is missing from that list and from the payout setup, OpenTrain AI cannot pay out there today, and there is no published timeline for adding new countries. The homepage separately says "Pay securely across 230+ countries."
• Approved money is hard to recall: released funds are not self-serve reversible in the platform, so mistaken approvals need a support case.
• Holds: OpenTrain may delay, hold or reverse payments or payouts for fraud prevention, compliance, disputes, chargebacks, legal requirements or risk management.
• Off-platform restriction: moving an introduced client or trainer to off-platform payment is barred during the engagement and for a 24-month tail period, unless OpenTrain permits it in writing or through a published opt-out.

External job listings

Jobs pulled in from outside platforms are not OpenTrain engagements. For those listings OpenTrain does not screen the applicant, hold the contract, run the engagement, hold payment or process payout. OpenTrain may receive a referral fee or commission when a referred user is hired there, and external listings may be stale, edited, withdrawn or unavailable by the time a user clicks through.

Self-reported figures and user feedback

Network figures vary across official pages: the Security page's pricing block says "Browse and hire from 100,000+ experts," against 511,000+ elsewhere, and the managed-service page counts native-speaker annotation across 118 languages while the homepage gives 120+ languages. All of these counts are vendor-reported.

As captured on October 5, 2026, a third-party review platform showed 3.9 from 109 reviews on a profile the company has claimed and pays a subscription for. The platform's AI-generated summary of 100 recent reviews is mostly positive about payouts and the admin team, but some reviewers found the AI screening steps frustrating, overly demanding and a source of unnecessary technical hurdles.

Data handling and privacy

• Model training: the Security page says customer datasets are never used to train third-party foundation models. Anthropic and OpenAI are listed as subprocessors for LLM inference for platform features only.
• Security controls: the Security page lists TLS 1.2+ in transit and AES-256 at rest for databases and object storage.
• Compliance documents: a Data Processing Addendum with Standard Contractual Clauses is available for GDPR and CCPA, shared with the CAIQ self-assessment under a mutual NDA; the page states there is no such thing as a government "HIPAA certification."
• Processor role differs by page: the FAQ says the client owns the data and OpenTrain is "a talent provider, not a data processor," while the Privacy Policy says OpenTrain may act as a processor when a customer uploads datasets containing personal data or when it manages a project under the customer's direction.
• Monitoring: communications on the Services may be monitored, reviewed and retained for safety, compliance, support and quality purposes.
• Public trainer profiles: a profile viewable without logging in can show first name and last initial, hourly rate and work stats including total earned; public profiles are currently set to noindex. Trainers can make a profile public, visible only to signed-in OpenTrain users, or private.
• Erasure: requests have a fulfillment target within 30 calendar days of verification, but transaction and tax records that must be kept by law remain.

Legal terms

Total liability is capped at the greater of fees paid to OpenTrain in the six months before a claim or US $100. Certain disputes may go to binding individual arbitration, with an opt-out by written notice within 30 days of first accepting the Terms, and the Terms are governed by the laws of the State of California.

FAQ

Q1. How much does OpenTrain AI cost for companies?

Posting is free; self-service adds 10% to talent payments plus $9.95 per new contract, and managed service charges 20% on top of the team's rate.

Q2. Do freelancers pay to find AI training jobs?

The freelancer page says no, although the Terms list talent-side fees as a possible fee type.

Q3. Are uploaded datasets used to train AI models?

The Security page says they are never used to train third-party foundation models.