Remote Data Labeling Jobs Paris

Rex.zone; United States; Remote; USD 41,000 - 69,000; Full time

Job summary

Rex.zone is hiring for a fully remote data labeling role based in the United States, focusing on training-data creation for AI/ML systems. You will label text, images, audio, and video to support RLHF, prompt evaluation, and content safety labeling.

You will follow annotation guidelines, document edge cases, and participate in calibration sessions to improve data quality and model performance. You will execute RLHF workflows, assess prompts for accuracy and safety, and maintain strict

Qualifications

• Experience in data labeling/annotation operations (text and/or computer vision).
• Familiarity with NLP tasks (e.g., named entity recognition, intent classification, semantic similarity).
• Ability to learn and apply strict rubrics for RLHF and QA evaluation.
• High attention to detail and consistency under changing guidelines.
• Comfort handling sensitive content in content safety labeling workflows.
• Reliable internet and availability for full-time remote work.

Responsibilities

• Perform data annotation across text, image, audio, and video tasks using production labeling tools.
• Execute RLHF workflows including pairwise preference ranking, justification tagging, and rubric-based scoring.
• Conduct prompt evaluation and response grading for accuracy, instruction-following, and safety.
• Apply content safety labeling policies (toxicity, self-harm, hate, sensitive attributes).
• Maintain annotation guidelines compliance; elevate ambiguous cases with clear examples.
• Participate in QA evaluation, audits, and disagreement resolution to improve training data quality.
• Track labeling throughput and accuracy metrics while meeting full-time delivery goals.
• Provide feedback to improve rubrics, taxonomies, and labeling ontology for scalable LLM training pipelines.

Skills

Data labeling

NLP tasks

RLHF rubrics

Attention to detail

Clear writing

Remote work

Job description

About The Role

Rex.zone is hiring for remote data labeling jobs aligned with Paris talent, focused on training-data creation for AI/ML systems. You will label text, images, audio, and video to support large language model evaluation, RLHF (Reinforcement Learning from Human Feedback), prompt evaluation, and content safety labeling. You will follow annotation guidelines, document edge cases, and contribute to calibration sessions to improve training data quality and model performance improvement.

What You Will Do

• Perform data annotation across text, image, audio, and video tasks using production labeling tools
• Execute RLHF workflows including pairwise preference ranking, justification tagging, and rubric-based scoring
• Conduct prompt evaluation and response grading for accuracy, instruction-following, and safety
• Apply content safety labeling policies (toxicity, self-harm, hate, sensitive attributes)
• Maintain annotation guidelines compliance; elevate ambiguous cases with clear examples
• Participate in QA evaluation, audits, and disagreement resolution to improve training data quality
• Track labeling throughput and accuracy metrics while meeting full-time delivery goals
• Provide feedback to improve rubrics, taxonomies, and labeling ontology for scalable LLM training pipelines

Required Qualifications

• Experience in data labeling/data annotation operations (text and/or computer vision annotation)
• Familiarity with NLP tasks (e.g., named entity recognition, intent classification, semantic similarity)
• Ability to learn and apply strict rubrics for RLHF and QA evaluation
• High attention to detail and consistency under changing guidelines
• Comfort handling sensitive content in content safety labeling workflows
• Clear written communication for edge-case documentation and reviewer notes
• Reliable internet and availability for full-time remote work

Compensation

Hourly range: $30–$50 USD (based on annotation experience, QA performance, and domain fit across NLP, computer vision, RLHF, and content safety).

Remote role managed under a US-based posting entity.