Remote Retrieval Relevance Evaluator — Quality Feedback

AuraOne; United States; On-site; USD 34,440 - 55,104; Full time

Job summary

AuraOne is seeking a Retrieval Relevance Task Evaluator to review prompts and responses against our quality rubric. This remote contractor role involves evaluating model outputs, labeling edge cases, and writing structured feedback to help retrain the system.

You will compare paired responses, assign severity tags, and document recurring failure modes. Reliable 10-hour weekly availability and strong attention to detail are essential to maintain calibration against gold-standard examples.

Qualifications

• Prior evaluation, annotation, or human-rater experience on retrieval relevance task evaluation.
• Comfort applying multi-page rubrics consistently across long batches.
• Clear written reasoning that names the issue and the rubric clause being applied.
• Strong attention to detail and the ability to flag when a prompt itself is the problem.
• Reliable async availability for at least 10 hours per week.

Responsibilities

• Evaluate retrieval relevance task evaluation model outputs against a versioned rubric and assign severity tags for 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

Evaluation experience

Rubric application

Written reasoning

Attention to detail

10 hrs/week availability

Job description

AuraOne is seeking a Retrieval Relevance Task Evaluator to review prompts and responses against our quality rubric. This remote contractor role involves evaluating model outputs, labeling edge cases, and writing structured feedback to help retrain the system.

You will compare paired responses, assign severity tags, and document recurring failure modes. Reliable 10-hour weekly availability and strong attention to detail are essential to maintain calibration against gold-standard examples.