Remote Scene Understanding Evaluator - AI Data Quality

AuraOne; United States; On-site; USD 34,000 - 62,000; Full time

Job summary

AuraOne is seeking a Remote Scene Understanding Evaluator contractor to review prompts and model outputs against our quality rubric. You will compare paired outputs, label edge cases, and write concise rationales that help the modeling team retrain.

The role focuses on identifying hallucinations, instruction-following failures, and unsafe content, with opportunities to calibrate against gold-standard examples each week.

Qualifications

• Prior evaluation, annotation, or human-rater experience on scene understanding 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.
• 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 scene understanding evaluation model outputs against a versioned rubric and assign severity tags for Scene Understanding 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.
• 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

Annotation

Written reasoning

Attention to detail

Async availability

Job description

AuraOne is seeking a Remote Scene Understanding Evaluator contractor to review prompts and model outputs against our quality rubric. You will compare paired outputs, label edge cases, and write concise rationales that help the modeling team retrain.

The role focuses on identifying hallucinations, instruction-following failures, and unsafe content, with opportunities to calibrate against gold-standard examples each week.