What You’ll Do
Team operations & performance: hire, onboard, run regular 1:1s and periodic reviews; drive throughput and consistency.
Labeling policy & taxonomy: maintain concise guidelines and edge-case libraries; turn complex rules into executable steps.
Pipeline & scheduling: balance labeling / review / sampling queues and daily priorities; ship versioned releases on time.
Quality & data governance: run multi-stage QC (review + sampling + gold sets), track accuracy/consistency/rework, ensure traceability.
Cross-functional: align hard cases and policy updates with ML; co-design sampling with Data Aug; collaborate with Engineering on tool UX/automation.
Compliance & vendors (if applicable): manage external providers, progress, and confidentiality.
Job Requirements
Proven experience managing a moderate-scale manual labeling team and delivering under changing priorities.
Hands-on background in image/video labeling; working grasp of 3D mesh / point cloud / depth / panorama data and QC pitfalls.
Ability to convert nuanced rules into clear SOPs/forms; strong documentation and communication.
Proficient with Web Data Annotation tools (or similar) and comfortable with basic data analysis (spreadsheets/SQL/scripts—any one).
Strong ownership for data accuracy, integrity, and timeliness; capable English reading/writing for work docs.
Preferred
Experience with city-scale mesh/point-cloud semantic segmentation; QGIS / CloudCompare / Blender for visual checks.
Understanding of geometric vision tasks and data needs (pose+depth, depth completion; Transformer/diffusion friendly datasets).
Basic Python for consistency checks and internal tooling.
Familiarity with data management and permissions; remote/multi-time-zone or vendor management experience.