A custom hardware interface on NVIDIA Jetson Orin Nano and Mac: a gesture-controlled interactive smart mirror with agentic capabilities.
Question
Generative models are fluent in text and image. They are not, by default, in the room. What does an agent look like when the interface is a surface you already have — a mirror — and the input is a hand in space?
Constraint
The stack has to close: camera, on-device vision, gesture, generative model, and a physical display that does not feel like a kiosk demo.
Method
A custom hardware interface using NVIDIA Jetson Orin Nano and Mac, with AI-driven computer vision processing real-time spatial inputs, bridging physical hardware and generative models on a single interactive surface.
Artifact
A working gesture-controlled smart mirror with agentic capabilities.
Engineered as a hardware–software research artifact, not a concept render. Live at maayaway.com.