AI Data Labeling Jobs Brazil

Rex.zone is hiring for remote, full-time AI Data Labeling roles focused on training data creation and evaluation workflows for AI/ML systems. Your work will support LLM training pipelines and multimodal model development through consistent, rubric-driven labeling and review.

Responsibilities

• Label text, image, and/or multimodal data following detailed annotation guidelines
• Perform RLHF tasks such as preference ranking and helpfulness/harmlessness evaluations
• Execute QA evaluation (spot checks, consensus review, error categorization) to improve training data quality
• Conduct prompt evaluation and rubric-based scoring for instruction tuning and LLM evaluation
• Document edge cases, propose clarifications, and contribute to feedback loops that improve model performance

Workstreams You May Support

• NLP labeling (intent, sentiment, toxicity, topic, summarization quality)
• Named entity recognition and entity linking
• Content safety labeling for policy compliance and risk mitigation
• Computer vision annotation (bounding boxes, polygons, keypoints)
• Grounding checks, reasoning quality ratings, and refusal/safety assessments

Requirements

• Experience in data annotation or QA evaluation in production workflows
• Strong annotation guideline compliance, accuracy, and consistency
• Comfort with structured labeling schemas and rubric-based scoring
• Ability to work independently in a remote environment and meet throughput targets

Preferred Qualifications

• Exposure to RLHF, prompt evaluation, LLM evaluation, NER, content safety, or computer vision annotation
• Familiarity with inter-annotator agreement, sampling-based QA, audits, and error analysis

Pay

Competitive hourly base pay: $30–$50 per hour (based on skills and project needs).