Remote Receipt & Invoice Quality Evaluator

AI Trainer Jobs; United States; Remote; USD 34,000 - 62,000; Full time

Job summary

AI Trainer Jobs is seeking a remote contractor to evaluate receipt and invoice understanding model outputs against AuraOne's rubric. You will review prompts, compare paired responses, label edge cases, and write structured feedback the modeling team can use to retrain.

As an independent specialist, you will assess model outputs, annotate with severity tags, and help calibrate inter-rater processes. This role requires reliable async availability (at least 10 hours/week) and meticulous attention

Qualifications

• Prior evaluation, annotation, or human-rater experience on receipt and invoice understanding model evaluation or adjacent content.
• Clear written reasoning that names the issue and the rubric clause being applied.
• Reliable async availability for at least 10 hours per week.
• Comfort applying multi-page rubrics consistently across long batches.

Responsibilities

• Evaluate receipt and invoice understanding model evaluation model outputs against a versioned rubric and assign severity tags for Receipt and Invoice Understanding Model Evaluator assignments.
• Compare paired responses and pick the stronger answer with a written rationale.
• Label hallucinations, instruction-following failures, and unsafe content with structured tags.
• Capture ambiguous prompts and route them back to the program team for rubric updates.

Skills

Attention to detail

Written reasoning

Reliable async availability

Comfort applying rubrics

Tools

Rubric-based annotation

Inter-rater calibration

Receipt and Invoice Understanding Model evaluation

Job description

AI Trainer Jobs is seeking a remote contractor to evaluate receipt and invoice understanding model outputs against AuraOne's rubric. You will review prompts, compare paired responses, label edge cases, and write structured feedback the modeling team can use to retrain.

As an independent specialist, you will assess model outputs, annotate with severity tags, and help calibrate inter-rater processes. This role requires reliable async availability (at least 10 hours/week) and meticulous attention