Scale has been the leading AI data foundry, helping fuel the most exciting advancements in AI, including frontier model training, enterprise adoption, defense applications, and autonomous vehicles. Our mission is to develop reliable AI systems for the world's most important decisions.

The next frontier for AI is the physical world. We're looking for an AI Product Manager to own the Robotics vertical within our Physical AI team. In this role, you'll own both the development of the data and training environments (the teleoperated demonstrations, real-world collections, simulated tasks, and annotation products that labs use to train and evaluate robot policies) and the "data as a product" strategy that powers them. You'll understand where physical AI is headed, decide what robot tasks and embodiments are worth collecting, how to source and structure the underlying data, and how to turn deep domain knowledge into a defensible product.

The ideal candidate has lived inside robotics or physical AI research, and is able to pair that domain understanding with a sense for where current robot policies succeed and fail in real-world workflows.

You'll translate that expertise into datasets, environments, and evaluation frameworks that teach robots to do real physical work, and you'll be the domain expert Scale's most important customers and their leading researchers turn to. A strong entrepreneurial & go-to-market mindset will be necessary.

What You'll Do

• Own the Robotics AI roadmap & data strategy: Set product direction for the robotics training stack and the data strategy behind it — what data we collect, on which hardware and embodiments, and what we source internally vs. through our marketplace. Establish a vision for where physical AI is heading, driving execution across engineering, operations, and go-to-market teams.
• Build partnerships with research teams at frontier labs: Work directly with researchers at leading physical AI labs to understand where their robot policies fall short and shape new product lines and competitive strategy for the vertical. Connect with robotics startups and industry leaders to launch joint benchmarks and build Scale's brand in physical AI.
• Design and scale robotics data products and environments: Scope and deliver high-quality collection pipelines, annotation tooling, simulated tasks, and evaluation frameworks — spanning robot-based collection (humanoids and robotic arms performing real-world tasks) and robot-less collection (wearable cameras capturing human demonstrations). Partner with ML and Operations to translate research needs into training products that hit quality, throughput, and cost targets.
• Collaborate cross-functionally — influence business priorities and dive into the weeds of research, operations, and customer interactions to deliver mission-critical outcomes. Travel ~10–15% to meet customers, attend conferences, and visit our global data operations.

Ideally, You'd Have

• 8 years of PM experience and 3 years of experience in robotics or physical AI in one or more of: robot learning or manipulation research, teleoperation and data collection systems, robotics software or hardware engineering, or a robotics product role, with real depth in how the work gets done.
• Physical AI Fluency: Immersed in current physical AI research — robot policies, vision-language-action models, language-conditioned imitation learning — and able to articulate how the field is moving, not just how classical robotics works.
• A builder's mindset: excited by ambiguity and motivated to create new products from the ground up.
• Product or customer-facing experience: A track record of owning outcomes, shaping roadmaps, or working closely with technical stakeholders (formal PM experience is a plus but not required if domain depth is strong).
• ML Intuition & Technical Fluency: Enough intuition around how model training and evaluation works, and what makes a dataset or environment actually useful for training. A software background is a plus.
• Operational rigor: Comfort going deep with Ops on quality and throughput, and reasoning about unit economics across hardware, labor, and annotation cost.
• Bias for action: Comfort wearing multiple hats and operating in fast-moving environments.
• A degree in Robotics or a related quantitative field (Computer Science, Mechanical Engineering, Electrical Engineering, etc.).

Nice to Have

• Ph.D. in Robotics or a related quantitative field, OR a Master's degree with 3+ years of equivalent professional experience in an applied research setting.