Product Manager, Physical AI Agents

About the role

You will own the definition and evolution of physical AI agents as a product category at Dexmate. You will decide how these agents behave, how developers interact with them, and what capabilities are exposed to the real world. You will translate the frontier of physical AI research into concrete product features that developers and enterprises can ship. You will set quality standards and evaluation frameworks that turn experimental robots into reliable production systems. You will act as the product voice across research, engineering, hardware, and early customers to ensure alignment on what agents can and should do. You will balance today's technical constraints with a vision of what the platform should become in six months and beyond.

Key facts

What you'll do

• Own the end to end roadmap for physical AI agents, from on robot policies and guardrails to the developer SDK and cloud evaluation and deployment surfaces.
• Translate advances in VLAs, imitation learning, reinforcement learning, and world models into features that developers can actually use to build reliable agents.
• Spend real time with developers, researchers, and enterprise users, observing how agents fail in the wild and identifying the highest leverage opportunities for the platform.
• Define what good looks like for physical AI agents by building evaluation frameworks, metrics, and review cadences covering success rate, recovery behavior, safety envelopes, and time to first skill.
• Shape the open ecosystem by deciding what to open source, what to publish, and how to cultivate a community of builders who accelerate physical AI development.
• Drive cross functional execution by aligning research, engineering, hardware, design, and go to market teams around a coherent product narrative and writing down tradeoffs.
• Defend product decisions with data, revisit them when evidence changes, and continuously update priorities based on what the platform reveals.
• Break down frontier capabilities into practical developer experiences that abstract complexity without limiting what is possible.
• Define and own the quality bar for agent behavior, including safety, reliability, and ease of integration into customer workflows.
• Act as the category owner for physical AI agents, setting the vision that makes the platform the default starting point for builders.

Requirements

• 5+ years of product management experience on technically deep products such as AI/ML, agents, developer platforms, robotics, autonomy, or similar domains.
• Track record of taking 0 to 1 products from research artifact to something people rely on in production environments.
• Strong AI agent sense with an opinionated, current view of where agents are going and what our roadmap should reflect.
• Deep technical fluency in the modern AI stack including VLAs, imitation learning, reinforcement learning, evaluation methodology, data pipelines, harness engineering, RAG, and tool use.
• Comfort at the hardware software boundary, understanding sensors, actuators, latency, failure modes, and the gap between simulation and the real world.
• Excellent product judgment under ambiguity, making crisp decisions with incomplete information and concise written communication.
• High agency and bias for shipping, delivering v1s quickly and iterating to quality instead of waiting for permission.
• Genuine love for builders, demonstrated by caring deeply about the experience of developers, researchers, and operators using the platform.

Nice to have

• Hands on technical background through engineering, applied research, or a computer science robotics graduate degree.
• Experience deploying and operating AI agents in real world environments.
• Contributions to open source robotics or AI agent projects.
• Familiarity with evaluation frameworks for agent quality and safety.
• Experience working with enterprise customers in hardware adjacent domains.

Practical notes

This role is based in the Fremont Office and requires full time onsite presence. Travel may be required to engage with users, partners, or research institutions. The position is exempt and eligible for standard employment authorization in the relevant jurisdiction. Visa sponsorship may be considered for qualified candidates where applicable.