Own machine learning models end to end, from data preparation and training through evaluation and production deployment. Build pipelines for large multimodal datasets including video, sensor streams, and language. Collaborate with engineering and computer vision teams, rapidly prototype solutions, apply relevant research, and continuously improve production models and tooling.

Conduct frontier research on robot learning and real-world training data. Design experiments with robotics, vision, and embodied AI models; study how data quality, structure, and diversity affect performance; develop evaluation methods; analyze relevant literature; and partner with engineering and data teams to turn findings into deployed pipeline and product improvements.