# OpenTrain AI

> OpenTrain is the open talent network for AI training and data labeling. AI teams can hire human evaluators, RLHF reviewers, red-teamers, annotators, and domain experts directly, or use OpenTrain managed service when they want the project run for them. Pricing is published: 10% self-service marketplace fee plus a $9.95 new-contract fee, and 20% managed service.

OpenTrain serves two sides of the AI training market:

- **For AI teams**: Post a project and hire AI Trainers and Data Labelers. Use self-service hiring or managed service, with talent working in the annotation stack you already use.
- **For AI Trainers**: Find AI training and data labeling projects across platforms, all in one place. Freelancer work history aggregates across 20+ platforms.

## OpenTrain Source Pack

Approved, machine-readable references for agents and answer engines:

- [App llms.txt](https://app.opentrain.ai/llms.txt): Agent registration, job drafts, publishing, proposals, hiring, contracts, milestones, messages, webhooks, credits, and human co-sign gates.
- [Docs llms.txt](https://www.opentrain.ai/docs/llms.txt): OpenTrain docs index for agents.
- [Docs llms-full.txt](https://www.opentrain.ai/docs/llms-full.txt): Expanded OpenTrain docs index for agents.
- [Jobs board llms.txt](https://www.opentrain.ai/jobs/llms.txt): Live, machine-readable directory of open AI-training and data-labeling jobs by category, subject-matter domain, country, and language, plus a sample of current roles with links.
- [Public OpenAPI spec](https://app.opentrain.ai/api/public/v1/openapi.json): Public v1 API contract.
- Official CLI package: `@opentrain-ai/cli`.
- Official MCP package: `@opentrain-ai/mcp`.

Hiring and money movement are human co-signed. An API, CLI, or MCP call can request a hire, fund a milestone, or release payment, but a signed-in human must confirm the money-moving action.

## Research Tools

OpenTrain provides two free research tools with structured, machine-readable data:

### Paper2Code Finder

Find open-source implementations for AI research papers. Each paper page provides:
- **Recommended implementation ranking** with confidence scores and maintenance status
- **Reproduction readiness** tier (ready_to_run / setup_required / major_work)
- **Benchmark numbers** tied to paper evidence (not hallucinated)
- **Time-to-reproduce** estimates (hours / days / unknown)
- **Hardware requirements** and risk flags
- **Hugging Face artifacts** (models, datasets, spaces)

**JSON API for AI agents:**
```
GET https://www.opentrain.ai/papers/{paperKey}.json
```
Returns structured JSON with implementation ranking, reproductionReadiness, benchmarks, riskFlags, and hardwareNotes.

Paper keys use the format `arxiv-{id}` (e.g., `arxiv-1706.03762` for "Attention Is All You Need").

**Example requests:**
- `https://www.opentrain.ai/papers/arxiv-1706.03762.json` — Transformer paper
- `https://www.opentrain.ai/papers/arxiv-2103.00020.json` — CLIP paper

**Search API:**
```
GET https://marketing-research-api.weston-114.workers.dev/api/search?q={query}&limit=8
```

**Resolve paper by arXiv ID, DOI, or title:**
```
POST https://marketing-research-api.weston-114.workers.dev/api/papers/resolve
Body: { "input": "2103.00020" }
```

### HF Eval Papers Explorer (HFEPX)

Browse and compare human evaluation protocols across AI research papers. Each paper includes:
- **Evaluation methodology extraction**: human feedback types, evaluation modes, quality controls
- **Benchmark and metric coverage**
- **Rater population and annotation unit details**
- **Confidence scoring** for extraction quality
- **Related papers** by protocol overlap

Browse at: `https://www.opentrain.ai/tools/hf-eval-papers/`

**Search API:**
```
GET https://marketing-research-api.weston-114.workers.dev/api/hfepx/search?q={query}&limit=8
```

### Machine-Readable Research Data

1. **Real-time repo status** — Stars, last push date, CI, Docker, archived status (checked regularly, not training-data stale)
2. **Curated implementation ranking** — Confidence-scored recommendations, not guesses
3. **Grounded benchmark numbers** — Extracted from paper tables and repo READMEs with evidence references
4. **Reproduction time estimates** — Hours vs. days vs. unknown, based on infrastructure analysis
5. **Cross-source artifact linking** — Paper → repos → HF models → datasets in one view

## Key Pages

- [Homepage](https://www.opentrain.ai/): Overview of the platform, talent network stats, and getting started
- [How It Works](https://www.opentrain.ai/how-it-works/): Step-by-step explanation of the hiring process
- [Pricing](https://www.opentrain.ai/pricing/): Pricing plans and feature comparison
- [Solutions Overview](https://www.opentrain.ai/solutions/): All AI training and data labeling solution categories
- [Integrations](https://www.opentrain.ai/integrations/): Supported annotation and labeling platforms
- [Managed Service](https://www.opentrain.ai/managed-service/): End-to-end managed programs for large or complex projects
- [Paper2Code Finder](https://www.opentrain.ai/papers/): Search for any paper by identifier or title
- [HF Eval Papers Explorer](https://www.opentrain.ai/tools/hf-eval-papers/): Browse evaluation protocols
- [AI Glossary](https://www.opentrain.ai/glossary/abductive-reasoning/): 439 AI and machine learning terms defined

## Jobs Board (for AI Trainers)

OpenTrain aggregates remote AI-training and data-labeling jobs from 20+ platforms into one board. Every role is remote, and candidates apply with a free OpenTrain account.

- [Browse all jobs](https://www.opentrain.ai/jobs/): The live job board of remote AI-training and data-labeling roles
- [Jobs by category](https://www.opentrain.ai/jobs/categories/): Generative AI & RLHF, coding & software, transcription and translation, image/video annotation, audio & speech, search evaluation, medical, legal & finance, data collection, writing, and general annotation
- [Jobs by subject-matter domain](https://www.opentrain.ai/jobs/domains/): Medicine, law, software engineering, finance, mathematics, languages, and 70+ more specialties
- [Jobs by location](https://www.opentrain.ai/jobs/locations/): Countries where roles are currently open
- [Live jobs directory for agents](https://www.opentrain.ai/jobs/llms.txt): Machine-readable index of open roles by facet plus a sample of current listings, updated continuously

## LLM & Agent Solutions

- [LLM Evaluation](https://www.opentrain.ai/solutions/llm-evaluation/): Golden datasets, pairwise ranking, rubric scoring, judge validation, and more
- [RLHF & Preference Data](https://www.opentrain.ai/solutions/rlhf-and-preference-data/): Pairwise ranking, multi-criteria scoring, rewrites, failure tagging, and more
- [Red Teaming](https://www.opentrain.ai/solutions/ai-red-teaming/): Jailbreaks, prompt injection, policy bypass, data leakage, and more
- [Hallucination Audits](https://www.opentrain.ai/solutions/hallucination-audits/): Claim verification, citation checking, severity scoring, and more
- [Supervised Fine-Tuning](https://www.opentrain.ai/solutions/supervised-fine-tuning/): Instruction data creation, domain verification, synthetic data QA, and more
- [Code Generation Review](https://www.opentrain.ai/solutions/code-generation-review/): Verification, RLHF ranking, bug detection, security review, and more
- [Function Calling](https://www.opentrain.ai/solutions/function-calling/): Tool selection, argument annotation, multi-step traces, error recovery, and more
- [Agent Simulations](https://www.opentrain.ai/solutions/agent-simulations/): Scenario design, gold trajectories, rollout evaluation, failure tagging, and more
- [Computer Use Agent Training](https://www.opentrain.ai/solutions/computer-use-agent-training/): Trajectory data and evaluation for browser and desktop agents
- [Reasoning Problem Creation](https://www.opentrain.ai/solutions/reasoning-problem-creation/): Original problems with step-by-step solutions in math, code, and more
- [Reasoning Step Verification](https://www.opentrain.ai/solutions/reasoning-step-verification/): Step-level correctness labels for training process reward models
- [RL Environment Design](https://www.opentrain.ai/solutions/rl-environment-design/): Reward functions, training scenarios, and environment QA for RL agents
- [Multilingual AI Training](https://www.opentrain.ai/solutions/multilingual-ai-training/): Native-speaker RLHF, evaluation, prompt writing, and cultural accuracy checking

## Data Labeling Solutions

- [Image Annotation](https://www.opentrain.ai/solutions/image-annotation/)
- [Video Annotation](https://www.opentrain.ai/solutions/video-annotation/)
- [Document Annotation](https://www.opentrain.ai/solutions/document-annotation/)
- [Segmentation](https://www.opentrain.ai/solutions/segmentation/)
- [Text Annotation](https://www.opentrain.ai/solutions/text-annotation/)
- [Medical Imaging Annotation](https://www.opentrain.ai/solutions/medical-imaging-annotation/)
- [Satellite Imagery Annotation](https://www.opentrain.ai/solutions/satellite-imagery-annotation/)
- [Point Cloud Annotation](https://www.opentrain.ai/solutions/point-cloud-annotation/)
- [Speech and Audio Labeling](https://www.opentrain.ai/solutions/speech-and-audio-labeling/)
- [Time Series Annotation](https://www.opentrain.ai/solutions/time-series-annotation/)

## Supported Integrations

Talent works in the customer's existing annotation or internal tools. Customer data stays with the customer. Supported integrations include:

- [Label Studio](https://www.opentrain.ai/integrations/label-studio/)
- [Roboflow](https://www.opentrain.ai/integrations/roboflow/)
- [AWS SageMaker](https://www.opentrain.ai/integrations/aws-sagemaker-groundtruth/)
- [CVAT](https://www.opentrain.ai/integrations/cvat/)
- [Scale Studio](https://www.opentrain.ai/integrations/scale-ai/)
- [SuperAnnotate](https://www.opentrain.ai/integrations/superannotate/)
- [Encord](https://www.opentrain.ai/integrations/encord/)
- [Argilla](https://www.opentrain.ai/integrations/argilla/)
- [Dataloop](https://www.opentrain.ai/integrations/dataloop/)
- [Kili Technology](https://www.opentrain.ai/integrations/kili-technology/)
- [Prodigy](https://www.opentrain.ai/integrations/prodigy/)
- [Snorkel AI](https://www.opentrain.ai/integrations/snorkel-ai/)
- [Supervisely](https://www.opentrain.ai/integrations/supervisely/)
- [Datasaur](https://www.opentrain.ai/integrations/datasaur/)
- [Datature](https://www.opentrain.ai/integrations/datature/)
- [Your Custom Tool](https://www.opentrain.ai/integrations/custom-tool/): Bring your own annotation platform

## Optional

- [Contact](https://www.opentrain.ai/contact/)
- [Get a Quote](https://www.opentrain.ai/get-a-quote/)
- [Become a Freelancer](https://www.opentrain.ai/become-freelancer/)
- [Privacy Policy](https://www.opentrain.ai/privacy/)
- [Terms of Service](https://www.opentrain.ai/terms-of-service/)