ETA Compute Product Design

In the Summer of 2021, I interned at ETA Compute, a startup focused on battery powered - AI vision sensors. I performed configuration studies to determine the device placement and design requirements.

To save on battery and device cost, the device used a low resolution, low fps camera. Because of this, it was crucial to optimize the device angle to maximize the number of clear frames the vision algorithm could use. I found that contrary to typical people counter devices, placement to the side of a door frame would maximize the number of frames a person would be seen. I was able to validate this choice by rendering walking animations from multiple camera angles and feeding those to the ML algorithm. I was also able to test multiple camera field of views, resolutions, and frame rates in this way.

I also designed and built the first physical prototype of ETA Compute's device. Several of these devices were placed at a partner's office in the products' initial launch in Fall 2021.