Founder · Technologist · Educator · Speaker
Today, at Labelbees, I'm building an operating platform for Physical AI. We're enabling robotics and AI teams to build, operate, evaluate, and continuously improve AI systems in the physical world.
Current Work
The operating platform for Physical AI.
Robotics and AI teams use Labelbees to build, operate, evaluate, and continuously improve AI systems as they move from the lab into the physical world — where sensors drift, edge cases multiply, and models have to keep learning long after deployment.
Production AI systems rarely fail because of a single model. More often, they fail because the systems around them break down, data drifts, feedback loops break down, evaluation becomes inconsistent, and real-world complexity compounds.
My work focuses on building the operational systems that enable AI to succeed in production, from multimodal data and model evaluation to verification, human-in-the-loop workflows, and continuous improvement.
Enabling robotics and AI teams to build, operate, evaluate, and continuously improve AI systems in the physical world.
Designing systems for model evaluation, verification, feedback loops, and continuous improvement.
Sharing practical lessons from building production AI systems through conferences, guest lectures, and industry events.
Track Record
Building reliable AI systems has been the common thread throughout my career, from founding startups to leading production AI engineering organizations across enterprise AI, Earth and space intelligence, and now Physical AI.
Built
Led At