Zippin's 2nd Checkout-Free Store — São Paulo, Brazil

Full Stack Software Engineer / Team Lead, Zippin — November 2019

Zippin builds checkout-free store technology: overhead cameras and shelf weight sensors identify what a shopper picks up, so they can walk in, grab what they want, and walk out with no scanning or line. This store was Zippin's second customer deployment, built into a large residential building in São Paulo.

I led the store opening from the Zippin side, spending two weeks on-site in Brazil installing and testing the in-store hardware. The store opened successfully, on schedule.

Packing

Boxes of NVIDIA Jetson Nano developer kits and SanDisk SD cards packed for shipment — Jetson Nanos, Raspberry Pis, and SD cards packed up for the trip.

Before the trip, I coordinated with the customer's import team on shipping, customs, and the bill of materials for all the hardware headed into Brazil.

Setup

On-site, my work consisted of flashing and configuring the edge devices that run the store, testing, validating, and calibrating the shelf weight sensors, and validating the camera setup.

Store under construction with shelving and camera mounts being installed — The store mid-construction — shelving and overhead camera mounts going in.

Raspberry Pis connected to a travel router in a hotel room for overnight flashing — Hotel room downtime: flashing multiple Raspberry Pis at once over a local network, since I only had so many laptop ports.

Raspberry Pis and a Jetson Nano being configured and boot-tested on a workbench — Configuring and boot-testing edge devices on-site, syncing with engineers back in California across time zones.

Wiring for shelf weight sensors behind a refrigerated display unit — Weight sensor wiring behind the shelving — taring and reliability were the tricky parts to get right on every unit.

Taking a lux (light level) measurement in the store with a handheld light meter — Measuring ambient light levels to validate lighting conditions for the overhead cameras.

Store Opening

Shoppers

What I did

Flashed and configured edge IoT devices — Raspberry Pis and NVIDIA Jetson Nanos — for the store's cameras and shelf sensors.
Communicated with the customer's import team on shipping, customs, and the bill of materials for the hardware shipment.
Used hotel room downtime and the room's router to set up a local network and flash multiple Raspberry Pis in parallel, working around limited laptop ports.
Coordinated setup progress and troubleshooting with engineers back in California across time zones.
Tested and calibrated the weight sensors on every shelf, including taring and working around sensor reliability issues.
Tested the camera setup and camera map/navigation configuration used to track shoppers and items through the store.
Measured ambient light levels (lux) around the store to validate conditions for the computer vision system.
Led the store opening on the Zippin side, on schedule.