WHAT YOU WILL BE WORKING ON

Main challenge

You’ll help build a high-quality computer vision dataset at scale (~1,000 real-world objects) that powers reliable object recognition.

The challenge is ensuring data consistency and quality across diverse real-world conditions like lighting changes, occlusion, motion blur, and reflections.

What it means on a daily basis

• Organize, clean, and structure large image/video datasets (naming, metadata, removing duplicates, filtering low-quality samples)
• Annotate data accurately and follow labeling standards to ensure consistency at scale
• Track dataset progress, coverage, and versioning (logs, changelogs, dataset updates)
• Perform quality checks and continuously improve dataset reliability through structured QA
• Work closely with AI engineers to refine data based on model performance and edge cases