Remote Data Labeling Specialist

About the Role

Rex.zone is seeking Data Labeling Specialists aligned with Munich timezone for remote work. This full-time position involves generating precise training and evaluation data for AI applications. The role encompasses labeling and reviewing datasets including text, images, and multimodal data used for language model training, reinforcement learning from human feedback (RLHF), prompt evaluation, and content safety assessments with structured quality assurance processes.

Key Responsibilities

• Annotate data for natural language processing and computer vision tasks, including classification, ranking, and multimodal labeling.
• Develop preference rankings in RLHF-style with transparent, rubric-based justifications.
• Conduct prompt assessments focusing on helpfulness, harmlessness, and overall quality metrics.
• Perform named entity recognition labeling within specialized domain datasets.
• Apply content safety labels following established policy categories and exercise sound judgment.
• Engage in quality assurance cycles including sample reviews, gold-set checks, resolving annotation disagreements, and calibration meetings.
• Identify ambiguous guidelines, record complex cases, and integrate updates to enhance labeling consistency.

Required Qualifications

• Hands-on experience in data labeling or quality assurance within a results-oriented environment.
• Exceptional attention to detail and strict adherence to annotation standards.
• Effective written communication and logical reasoning skills for ranking and evaluation tasks.
• Capability to work independently and asynchronously in a distributed remote team setting.

Preferred Skills and Experience

• Knowledge of RLHF methodologies, large language model evaluation, or dataset creation for model assessments.
• Familiarity with content safety labeling guidelines and calibration techniques.
• Experience using computer vision annotation platforms or participating in named entity recognition projects.

Additional Information

This opportunity requires candidates to complete a short screening process, a guideline-driven assessment, and a quality assurance calibration step before project assignments supporting LLM training pipelines.