Computer Vision Annotator (Remote)

About the role

The Computer Vision Annotator works remotely to create high‑quality labeled datasets that power computer‑vision and machine‑learning models. You will use specialized annotation tools to mark objects, key points, 3D structures and pixel‑level masks across images and videos.

Key responsibilities

• Draw precise bounding boxes, polygons, polylines and 3D cuboids around objects of interest.
• Place key points for pose estimation and annotate point‑cloud data for LiDAR or robotics applications.
• Perform semantic and instance segmentation, labeling every pixel in an image.
• Interpret complex labeling taxonomies and refine guidelines in collaboration with project managers and data scientists.
• Utilize advanced features of tools such as CVAT, Label Studio, 3D Slicer, Supervisely or Amazon SageMaker Ground Truth.
• Identify ambiguous cases, edge‑cases and potential biases, and report findings to improve model robustness.
• Contribute to best‑practice documentation, workflow optimisations and training material for new annotators.
• Assist with basic data preparation tasks like image resizing, format conversion and sensor data synchronization.

Required profile

• Minimum 2 years of hands‑on experience in data annotation or quality assurance for computer‑vision projects.
• Exceptional attention to detail and a strong commitment to accuracy and consistency.
• Excellent written and verbal communication skills in English.
• Ability to document workflows clearly and collaborate effectively with remote teams.

Required skills

• Proficiency with annotation tools: CVAT, Label Studio, 3D Slicer, Supervisely, Amazon SageMaker Ground Truth.
• Experience with bounding‑box, polygon, polyline, key‑point, 3D cuboid and segmentation‑mask techniques.
• Solid understanding of computer‑vision concepts such as object detection, image classification, LiDAR and point clouds.
• Familiarity with quality‑control processes like consensus labeling, audit sampling and inter‑annotator agreement metrics.