Vice President of Embodied AI

We are seeking a Vice President of Embodied AI to lead the "brain" and "nervous system" of our humanoid robots. This role owns the company's AI strategy and the core AI and controls organization — real-time controls, whole-body reinforcement learning, dexterous manipulation, VLA post-training, and agentic autonomy — with a primary mandate to unify these stacks into a single, coherent system that performs reliable, useful work for our customers.

You will lead a world-class, multi-disciplinary organization of 45+ engineers and researchers. You will own the overarching technical AI strategy and data collection strategy, and serve as a crucial voice in executive discussions on the product and technology roadmap.

We are looking for a unique hybrid: a recognized thought leader in the embodied AI research community who possesses the hard-won pragmatism and engineering rigor required to ship physical products to commercial customers.

• Define and own the company's embodied AI vision and roadmap, aligned with product strategy and long-term autonomy goals.
• Drive technical decision-making around model architectures and the role of internal models versus strategic partner and third-party foundation and VLA models.
• Establish architectural principles for how learned and classical components compose into a safe, reliable robotic system.
• Partner with the CEO, CTO, and executive team to define the product and technology roadmap, translating complex AI capabilities into commercial milestones.

• Architect and execute a unified controls strategy, bridging high-level semantic reasoning (VLAs and agentic planning), mid-level policy execution (RL), and low-level deterministic real-time control.
• Define clean interfaces, arbitration, and fallback behavior across the layers of the stack so that capability gains in one layer compound rather than conflict.
• Ensure the unified stack meets the latency, stability, and safety requirements of dynamic humanoid platforms operating around people.

• Lead development of the models and policies that power the robot: state estimation and real-time control, whole-body RL locomotion and coordination, dexterous high-DoF manipulation, VLA post-training and adaptation, and agentic task planning and error recovery.
• Establish best practices for fine-tuning, distillation and compression, safety constraints and guardrails, and continuous learning.
• Define the evaluation criteria and benchmarks that tie model performance to real-world robotic outcomes.
• Guide the adaptation of strategic partner and third-party foundation models into the production stack, prioritizing co-optimization where it drives performance, safety, and speed to deployment.

• Own the end-to-end data strategy required to train state-of-the-art embodied AI: what data to collect, from which sources, at what scale, and to what quality bar.
• Define and drive data collection programs across teleoperation, real-world robot fleets, and synthetic data and simulation.
• Set dataset requirements, curation standards, and quality metrics for all training and evaluation data.
• Direct the design of simulation environments and scenarios used for training, evaluation, and sim-to-real transfer.

• Ensure our controllers are reliable, safe, and deliver tangible ROI for customers.
• Own the definition of model readiness for deployment — the performance, safety, and robustness criteria a model must meet before reaching customer sites.
• Use field telemetry and deployment feedback to close the loop between real-world behavior, data collection priorities, and model improvement.

• Act as an ambassador for the company within the global AI and robotics communities.
• Publicize key findings where aligned with IP strategy, and build strategic research and industry relationships.
• Attract, hire, and retain top-tier engineering and research talent.

• Directly manage and scale a multi-disciplinary organization of engineers and researchers across real-time controls, whole-body RL control, dexterous hand control, VLA post-training, and autonomy frameworks.
• Collaborate with the software, infrastructure, and hardware organizations to ensure the AI stack is well supported from training through on-robot deployment.
• Set engineering and research standards, review practices, and career development paths.
• Foster a culture of technical excellence, experimentation, accountability, and cross-functional collaboration.

• 10+ years in robotics, AI, or machine learning, with 5+ years in senior leadership (VP/Director) managing large, multi-disciplinary technical organizations (40+ engineers/researchers).
• Deep technical fluency across the modern robotics stack, including the trade-offs between classical control theory, reinforcement learning, and modern foundation models (VLAs, world action models, LLMs).
• Capable of taking complex, AI-driven hardware or robotics systems out of R&D and successfully deploying them to external customers in the real world — balancing "perfect" research with "good enough to ship."
• Experience defining and scaling the data strategy behind large-scale action models: teleoperation, real-world collection, auto-labeling, and sim-to-real transfer.
• Respected presence in the AI/robotics research community (e.g., publications at ICRA, IROS, CoRL, NeurIPS, CVPR), with a network that supports strategic hiring and partnerships.
• Exceptional communication: able to distill complex technical constraints into clear strategic decisions for the executive team, while diving deep into architecture discussions with staff engineers.

• Direct experience with humanoid or legged robotics, dexterous manipulation, or dynamic whole-body control.
• Experience integrating and co-optimizing with strategic partners or third-party foundation and VLA models.
• Experience navigating the safety and compliance challenges of deploying autonomous robots in human-centric environments.
• Advanced degree in Robotics, Computer Science, ML, or a related field.

• The controls and AI stacks operate as one unified architecture, from semantic reasoning down to real-time actuation.
• Robots reliably perform useful, revenue-generating work for customers, with AI capability translating directly into commercial milestones.
• The data collection strategy measurably accelerates model improvement across teleoperation, fleet, and simulation sources.
• Internal, partner, and third-party models are adopted pragmatically, based on what delivers performance, safety, and speed to deployment.
• The organization attracts world-class talent and is recognized as a leader in applied embodied AI, not just research.
• Executive and product decisions are grounded in a clear-eyed view of what the AI stack can deliver and when.

• Prolonged periods of sitting at a desk and working on a computer
• Must be able to lift 15 pounds at times
• Vision to read printed materials and a computer screen
• Hearing and speech to communicate