Work

I work on how people control robots. That covers teleoperation interfaces — the systems an operator uses to drive a robot directly — along with the infrastructure those interfaces run on and the data collection pipelines that turn operator sessions into training data.

Teleoperation sits in an unusual place. It is a real-time control problem, a UX problem, and a data problem at once: the same session that lets a person complete a task is also the demonstration a model learns from. Designing for the operator and designing for the dataset are not the same thing, and most of the interesting decisions live in that tension.

This work supports DeepMind's broader robotics effort, including Gemini Robotics — vision-language-action models intended as a general intelligence layer for robots, able to generalize across different hardware and to tasks they were not explicitly trained on, rather than being purpose-built per task.

Robotics interface work within Google's Brain team, before the org became part of Google DeepMind.

I worked on Project Reach, whose goal was to democratize remote robotic teleoperation — making a wide variety of robotic arms operable from a distance, rather than each platform needing its own bespoke control stack. Most of that problem is finding an abstraction general enough to cover hardware that disagrees about nearly everything, without flattening away the differences that actually matter to an operator.

I also built teleoperation for MetA, the Everyday Robots mobile manipulator platform, supporting both continuous control — an operator driving the arm directly in real time — and action-scripted control, where predefined behaviors are composed and dispatched.

Unity contract engineer, working on-site with engineering teams at Google and at an automotive manufacturer's research institute. Built games, tools, and AR/VR applications — UX research in VR, AR SDK support, and robotics UI.

Built games and interactive experiences in Unity to market and demonstrate client technologies. Worked across teams of 2D and 3D artists and programmers on web content, mobile games, and VR experiences, primarily showcased at security and big data conferences.