• Design Physical AI problem formulations that map real‑world robotic behavior into concrete data and evaluation requirements for training policies, world models, and perception systems. • Prototype perception, world‑model, and action‑representation pipelines (e.g., VLMs, VLAs, world models) to understand what data is needed, why it matters, and how quality will be measured. • Use simulation and synthetic environments (digital twins, Isaac/Omniverse‑style tools) to generate, stress‑test, and scale datasets for robotics and humanoid systems, grounded in real sensors, tasks, and constraints. • Work directly with customers’ robotics and ML teams to define data specifications, collection strategies (egocentric capture, teacher‑follower demonstrations, imitation learning), and evaluation benchmarks that tie to model performance and business outcomes. • Lead technical discovery and pre‑sales pilots: scope projects, design experiments, and secure the “technical win” by demonstrating uplift from our data, annotations, and pipelines. • Collaborate with internal data‑collection and platform teams to design robust data pipelines, annotation workflows (including affordances and advanced CV labels), and QA processes that generalize across customers. • Develop reusable playbooks, reference architectures, and demos for common Physical‑AI use cases (manipulation, mobile navigation, teleoperation, human‑robot interaction) to accelerate future engagements. • Influence the product and tooling roadmap by bringing structured feedback from frontier robotics customers, and help shape a scalable “Robotics & Physical AI data platform.” • Represent the company at key industry events and workshops, evangelizing best practices for robotics data, simulation, and evaluation and helping build a broader data and partner ecosystem.