Document and Email Data Extraction Specialist

About OpenTrain

OpenTrain is the #1 platform for finding and building careers in AI training and data labeling. OpenTrain connects people with real projects that teach AI how to read, reason, and understand the world — create a free account, build a profile, and apply in minutes.

This listing is for AI training work: the human side of building modern models. Contributors help shape how AI systems behave by reviewing, correcting, and validating data that models learn from.

About AI Training Work

AI training (also called data labeling or human feedback) is a fast-growing, remote-friendly way to work with flexible hours. Tasks range from transcribing audio and annotating images to evaluating model outputs and validating structured data extracted from documents.

This role focuses on document-level validation: you will directly improve the quality of structured records that feed downstream AI systems used by clients in commodities and logistics.

The Role

You will review every structured record produced by our AI extraction system and compare it carefully to the source document (PDFs, emails, scanned images). Records contain fields such as dates, amounts, names, codes, and more — the exact schema changes by document type.

This is a part-time contractor role, remote and worldwide, with a target workload of less than 20 hours per week. Pay is hourly at USD 12.00.

What You'll Do

Work through tasks that contain source documents and the AI-extracted record. For each field in the record you will:

Handle a variety of document types — invoices, contracts, bills of lading, certificates, sales orders, purchase orders, and similar back-office or middle-office documents — each with its own field schema and descriptions.

• Confirm fields the AI extracted correctly by matching them to the source document.
• Correct fields that are incorrect and enter the accurate values.
• Fill in fields the AI missed when the information is present in the source.
• Flag or escalate anything ambiguous, missing, or unclear following provided escalation guidelines.
• Carefully read field names and descriptions each task; do not assume fields have the same meaning across different document types.

Requirements

We’re looking for intermediate-level contributors who can work accurately and consistently on structured data tasks. You must follow guidelines closely and make judgment calls when information is ambiguous.

The project uses a custom/third-party labeling interface (listed as OTHER). You must be comfortable learning and using new annotation tools.

• Experience working with back-office or middle-office documents (invoices, sales orders, purchase orders, bills of lading) is ideal.
• Prior data-labeling or review experience — especially with LLM products or structured-data extraction — is a strong plus.
• Attention to detail, good reading comprehension, and the ability to escalate ambiguous items appropriately.
• Available to work remotely and complete under 20 hours per week as a contractor.

Who Should Apply

Apply if you want flexible, remote work that directly improves AI systems and you have familiarity with business documents used in commodities and logistics.

This role fits people looking for part-time contractor work and who can reliably follow instructions and annotation schemas across varied document types.

How It Works & Compensation

You will receive batches of tasks via the labeling platform, review documents and extracted records, and submit validated records or corrections. Follow the provided guidelines and escalation process for ambiguous items.

Compensation is hourly at USD 12.00 per hour. Employment type: contractor, part-time. The project classification: document data extraction and evaluation (label types: evaluation_rating and data_collection).