AI Data Labeler

Mashgin powers the world's best checkout experience with AI technology, serving over 40 million users and handling over 1 billion transactions at various high-profile locations. The company is profitable, innovative, and culture-driven, focusing on creating impactful products.

Location: United States

Responsibilities

• Label images and video frames with bounding boxes, polygons, segmentation masks, and SKU-level classifications.
• Annotate edge cases, including occlusion, overlapping items, glare, motion blur, and unusual product orientations.
• Maintain consistency against the labeling taxonomy and follow detailed annotation guidelines.
• Tag training, validation, and test data to support model development and evaluation.
• Compare model predictions to ground-truth labels and document failure modes.
• Audit annotations from peers and contractors to enforce inter-annotator agreement.
• Flag systemic issues such as recurring misclassifications, mislabeled SKUs, or low-quality captures.
• Review confusion matrices and error reports with the ML team to prioritize fixes.
• Identify capture issues that indicate hardware problems: blurry frames, poor lighting, color shifts, dropped frames, or camera misalignment.
• Test devices in lab and field conditions to confirm image quality and end-to-end checkout accuracy.
• Reproduce and document software bugs surfaced by labeling workflows or production telemetry.
• Partner with hardware and software engineers to validate fixes and run regression checks.
• Maintain and refine internal labeling guidelines as new SKUs, packaging, and edge cases emerge.
• Write concise reports summarizing labeling trends, error patterns, and recommendations.
• Collaborate cross-functionally with ML engineers, hardware engineers, product, and operations.

Requirements

• 2+ years of experience labeling data for computer vision, robotics, autonomous vehicles, or medical imaging.
• Exceptional attention to detail and high tolerance for repetitive, precision-oriented work.
• Experience with annotation tools such as CVAT, Labelbox, SuperAnnotate, Scale, or in-house tooling.
• Comfort following detailed written guidelines and documenting ambiguous cases instead of guessing.
• Strong written communication for clear, structured QA reports and Slack updates.
• Comfort working with images and video from physical devices, and reasoning about visual edge cases.