Technical Program Manager, Dataset Operations

Details

We are looking for a Technical Program Manager (TPM) to lead the end-to-end management of large-scale, multi-modal datasets that power next-generation Physical AI systems. This role focuses on vendor-sourced data pipelines, ensuring data quality, compliance, scalability, and alignment with research and product goals. You will operate at the intersection of research, engineering, data vendors, and legal/compliance, driving execution across complex, high-stakes data programs.

Responsibilities

• Own the lifecycle of large-scale datasets across modalities, including:
  • Ego-centric video (AR/VR, human interaction)
  • Robotics data (manipulation, embodied AI)
  • Autonomous driving data (multi-sensor, multi-agent)
• Manage external data vendors end-to-end:
  • Scope definition, onboarding, and contracting
  • Data specification and annotation guidelines
  • Delivery tracking, quality control, and iteration loops
• Establish scalable processes for multi-vendor coordination
• Translate research and model requirements into clear, enforceable data specifications
• Define and track data quality metrics (coverage, diversity, labeling accuracy, temporal consistency)
• Drive continuous improvement via structured feedback loops with vendors
• Partner with research teams (to understand evolving model needs), engineering teams (data pipelines, storage, tooling), and legal/compliance (data usage rights, privacy, licensing)
• Align dataset strategy with model training and product timelines
• Build frameworks for dataset versioning and traceability, vendor performance benchmarking, and cost vs. quality optimization
• Drive automation where possible while maintaining strong operational rigor

Requirements

• 5+ years of experience as a TPM, Product Ops, or similar role in AI/ML or data-intensive systems
• Master's degree (or equivalent experience)
• Proven experience managing external vendors or large-scale data programs
• Strong understanding of ML data pipelines and dataset lifecycle
• Ability to operate in ambiguous, fast-moving research environments
• Familiarity with one or more of: computer vision / video datasets, robotics / embodied AI data, autonomous driving datasets
• Understanding of annotation workflows, data quality evaluation, dataset biases and coverage challenges
• Strong program structuring skills (clear specs, milestones, tracking) and ability to convert high-level goals into concrete execution plans
• Comfortable working across research, engineering, and external partners

Ways to Stand Out

• Experience with multi-modal datasets (video, sensor, action data)
• Background in robotics, self-driving, or AR/VR
• Experience building vendor ecosystems from scratch
• Familiarity with data compliance (GDPR, privacy, licensing constraints)

Why This Role Matters

High-quality data is the foundation of Physical AI. As models evolve toward world understanding and action, dataset complexity increases dramatically. This role is critical to ensuring reliable acquisition, management, and scaling of the data needed to train next-generation systems. The team builds systems to train NVIDIA's world foundation model for physical AI (Cosmos) enabling large-scale AI models for robots, autonomous agents, and AI systems to understand, plan, and act in complex environments.

Compensation & Additional Details

• Base salary ranges by level (location- and experience-dependent):
  • Level 4: 168,000 USD - 258,750 USD
  • Level 5: 200,000 USD - 322,000 USD
• Eligible for equity and company benefits
• Applications accepted at least until May 21, 2026

Company & Hiring Notes

• NVIDIA uses AI tools in its recruiting processes
• NVIDIA is an equal opportunity employer and values diversity in hiring and promotion practices