telus-digital - Physical AI Robotics Product Manager
Requirements
• 4+ years in product management, including at least 2 in robotics, physical AI, AR/VR, autonomous systems, or data collection at scale. • Technical degree, or equivalent experience. Past work as a software engineer or researcher is a plus. • Technical depth to hold a real conversation with a research team. • Direct experience with one or more of: egocentric or wearable data capture, teleoperation systems, robot data pipelines, multi-sensor collection rigs or similar setups • Experience defining product vision and roadmaps in ambiguous, fast-moving environments. • Hands-on robotics background. Robotics, ML engineer on embodied models, teleoperation engineer, or similar. • Familiarity with imitation learning, VLA models, or manipulation policy training, and the data those methods need. • Experience managing distributed human operations, including crowd or contractor workforces. • You have shipped something physical and understand calibration, sensor sync, and coordinate frames, and how they fail.
Responsibilities
• Own the roadmap for robotics data collection, setting direction and driving execution across engineering, design, operations, quality, and solutions. • Assess demand for new capture modalities, then define the collection model, hardware, and tooling. • Work with engineering and design on the operator experience: onboarding, task guidance, in-session feedback, error states, review workflows. Operators work one-handed, in motion, wearing a headset, or against a clock. • Build prototypes and write the specs. Requirements and user stories that hold up under engineering review. • Translate customer model requirements into collection designs, and evolving research into quality standards for egocentric data: pose accuracy, sync tolerance, occlusion handling, annotation schema, acceptance criteria. • Spend time with collectors, session leads, and reviewers. When data underperforms in training, trace it back to the capture design or the tooling that produced it. • Represent the product directly to hyperscaler and robotics foundation model teams, and bring their constraints back into the roadmap. • Optimize for yield per session, rejection rate, cost per usable hour, capture-to-delivery time, and operator retention. • Champion pilots as the default way to test a change. Small cohort, clear hypothesis, fixed window. Scale what works and stop what does not.
Benefits
• Early market: Standards for physical AI data are not set yet. You will have a hand in setting them. • Impact at scale: Hundreds of devices in the field and an operator base large enough that a design change shows up in the numbers within a week. • Hard/Unsolved problems: Capture design, quality definition, and operator experience for a data type most of the industry cannot produce.
Apply in one click
Upload My Resume
Drop here or click to browse · Tap to choose · PDF, DOCX, DOC, RTF, TXT