Data Annotation Specialist - Computer Vision

Role overview

Create reliable training data by annotating construction drawings for computer vision models. The work combines careful visual labeling with dataset cleanup and close feedback to machine learning engineers; guidelines may evolve as new cases arise.

Responsibilities - Mark drawing elements with bounding boxes, polygons, and segmentation masks according to accuracy standards. - Apply annotation rules consistently and flag unclear or unusual cases for clarification. - Correct existing datasets, including inconsistent labels, missing metadata, and duplicate records. - Meet weekly quality and production targets while maintaining accuracy during repetitive work. - Share observations about data patterns and model failures with engineering colleagues.

Requirements - Strong attention to detail, including the ability to catch omissions and inconsistencies independently. - Ability to follow written procedures closely and use sound judgment when instructions do not cover a case. - Sustained focus on repetitive tasks without a decline in accuracy. - Ability to learn new tools quickly and work independently in a remote setting. - Clear written English for documenting issues and asking questions.

Nice to have - Experience with image annotation tools such as Labelbox, CVAT, or Supervisely. - Familiarity with object detection and image segmentation concepts. - Ability to interpret construction, CAD, or architectural drawings.

Benefits and work setup

This is a remote U.S. contract expected to last three to six months, at 40 hours per week, with availability during core Pacific Time hours. The stated pay range is $20–$35 per hour, with performance bonuses; strong performers may have a path to full-time work. The source also describes equity as part of the compensation approach, with packages varying by role, experience, and contribution.