Remote Helpfulness Ranking Data Reviewer

AuraOne, Inc.; United States; Remote; USD 34,000 - 55,000; Full time

Benefits offered by this job

Remote work

Job summary

AuraOne, Inc. is seeking a Helpfulness Ranking Preference Data Reviewer for a remote contract-like role. You will evaluate model outputs against a versioned rubric, label edge cases, and provide structured feedback to retrain the system.

You will compare paired responses, select the stronger answer with a written rationale, and tag content issues such as hallucinations and unsafe content. This role requires strong attention to detail and reliable weekly calibration against gold standards.

Qualifications

• Prior evaluation, annotation, or human-rater experience on helpfulness ranking preference data evaluation or adjacent content.
• Comfort applying multi-page rubrics consistently across long batches.
• Clear written reasoning that names the issue and the rubric clause being applied.

Responsibilities

• Evaluate helpfulness ranking preference data evaluation model outputs against a versioned rubric and assign severity tags for reviewer 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.
• Maintain reviewer-quality scores by calibrating against gold-standard examples each week.
• Document recurring failure modes so the modeling team can target them in the next training run.

Skills

Model output evaluation

Rubric-based annotation

Severity tagging

Inter-rater calibration

Helpfulness Ranking Preference Data

Preference ranking

RLHF

Rater calibration

Attention to detail

Critical thinking

Job description

AuraOne, Inc. is seeking a Helpfulness Ranking Preference Data Reviewer for a remote contract-like role. You will evaluate model outputs against a versioned rubric, label edge cases, and provide structured feedback to retrain the system.

You will compare paired responses, select the stronger answer with a written rationale, and tag content issues such as hallucinations and unsafe content. This role requires strong attention to detail and reliable weekly calibration against gold standards.