Remote Data Labeling Boston

Rex.zone is hiring for a full-time, fully remote role focused on creating and evaluating high-quality training data for AI/ML systems. You will label text, images, audio, and video while following strict annotation guidelines compliance to improve training data quality for LLM training pipelines.

• Annotate and label text/image/audio/video samples for AI/ML datasets
• Perform QA evaluation, self-audits, and consistency checks (gold tasks, audits, inter-annotator agreement)
• Complete RLHF tasks such as pairwise preference ranking and rubric-based scoring with concise rationales
• Run prompt evaluation for helpfulness, harmlessness, and factuality to support large language model evaluation
• Label named entity recognition (NER) and relations for NLP tasks
• Support computer vision annotation (bounding boxes, polygons, segmentation/keypoints) where applicable
• Apply content safety labeling policies and elevate edge cases with clear examples
• Document decision rules to resolve ambiguity and improve batch-to-batch consistency

• Mid-Senior experience delivering accurate annotations at production scale
• Strong attention to detail and ability to follow strict taxonomies and policy requirements
• Comfort working independently in a remote, full-time setting while meeting quality and throughput goals
• Understanding of how labeling decisions impact model performance improvement and downstream metrics

This role is listed for Boston search intent; work is performed remotely within the US with reliable internet and secure workspace practices.

Hourly base pay range: $30-$50 per hour (based on skills alignment, dataset domain complexity, and QA evaluation performance).