Senior Manager, Product - Perception Data

NVIDIA NVIDIA · Semiconductors · Santa Clara, CA

Senior Manager of Product to lead a data team focused on perception models for the Automotive industry. The role involves understanding model needs, defining annotation strategies, owning the data roadmap, and driving delivery across labeling vendors and internal teams. Requires product management experience with ML/AI data pipelines and perception fundamentals.

What you'd actually do

  1. Develop a deep understanding of how training data impacts perception model performance across tasks including 3D object detection, lane/road structure recognition, traffic sign and traffic light detection, and semantic segmentation.
  2. Partner with perception researchers and engineers to translate model capability gaps into concrete data requirements.
  3. Own annotation ontology design: define labeling taxonomies, attribute schemas, edge case handling rules, and inter-annotator consistency standards for perception network tasks such as 3D bounding boxes, lane elements, traffic sign/light classification and association.
  4. Anticipate how changes in data density, ROI size, and label complexity affect labeling throughput and delivery capacity; provide data-backed forecasts to stakeholders.
  5. Collaborate with global engineering and labeling teams to ensure data quality and explore how DNN/LLM/VLM improves the performance of data labeling.

Skills

Required

  • BS degree in Computer Science, Engineering, Data Science, or related field (or equivalent experience)
  • 8+ years of product management experience
  • 2+ years in a role directly involving ML/AI model development, data pipelines, or training data strategy
  • Strong cross-functional communication skills
  • 3+ years of experience leading or mentoring a team of product managers, tech leads, or equivalent

Nice to have

  • Fluent in Mandarin Chinese
  • Direct experience in autonomous driving, robotics, or ADAS product development
  • Familiarity with 3D reconstruction, SLAM, NeRF, LiDAR point cloud annotation, and multi-sensor fusion
  • Exposure to model-in-the-loop, human-in-the-loop, auto labeling, or VLM labeling

What the JD emphasized

  • directly drive model performance gains
  • genuine technical depth in ML data pipelines and perception tasks
  • Genuine understanding of perception model fundamentals

Other signals

  • driving model performance gains
  • data roadmap
  • end-to-end delivery