Remote AI Data Annotator — NLP & Computer Vision

Rex.zone; Los Angeles (CA); Remote; USD 45,000 - 60,000; Full time

Job summary

Rex.zone is seeking mid-senior remote data annotators in a fully remote capacity to create and evaluate training data for AI/ML systems, including NLP, computer vision, and content safety workflows. You will follow detailed guidelines, complete calibrations, and contribute to high-quality datasets used in LLM training pipelines.

Responsibilities include applying labeling rules, performing RLHF tasks, and conducting QA evaluations while maintaining data quality under throughput targets.

Qualifications

• Experience in data annotation or QA evaluation.
• Familiarity with LLM evaluation concepts and rubrics.
• Ability to maintain data quality under throughput targets.

Responsibilities

• Apply data labeling rules to text, image, audio, or multimodal samples.
• Perform RLHF tasks (preference ranking, pairwise comparisons, rubric-based scoring)
• Conduct QA evaluation through self-checks, peer review, and targeted audits
• Complete NLP labeling such as named entity recognition and related taxonomy tasks
• Support prompt evaluation (relevance, factuality, instruction-following, safety)
• Complete computer vision annotation (bounding boxes/segmentation) when assigned
• Document edge cases and propose guideline clarifications; follow versioned specs
• Track annotation guidelines compliance and elevate ambiguous examples

Skills

Data labeling

QA evaluation

Throughput targets

Clear communication

Tools

QA trackers

Labeling tools

Job description

Rex.zone is seeking mid-senior remote data annotators in a fully remote capacity to create and evaluate training data for AI/ML systems, including NLP, computer vision, and content safety workflows. You will follow detailed guidelines, complete calibrations, and contribute to high-quality datasets used in LLM training pipelines.

Responsibilities include applying labeling rules, performing RLHF tasks, and conducting QA evaluations while maintaining data quality under throughput targets.