Abhilash Reddy Sirigi ReddyAbhilash Reddy

Work

AR in the aisles and faster scanning at Walmart

Android Engineer, AR and barcode scanning (contract), Walmart, 2025 to 2026

  • Kotlin
  • Android SDK
  • ARCore
  • Jetpack Compose
  • ML Kit
  • Zebra SDK (C++/JNI)
  • Room

Context

I came back to Walmart on a part-time contract. The work ran on two tracks: an augmented reality proof of concept built on ARCore, and production barcode scanning in the existing Walmart app.

The problem

In a distribution centre or a back office, a picker standing in an aisle has to work out what each product is, how many there should be and where it belongs. That information exists, but it is on another screen, not on the shelf in front of them.

The question for the AR track was whether that information could be shown on top of the product itself, on the fly, on the handheld devices workers already carry. The longer aim was a customer feature: tap a product in the app and be shown the way to it inside the store.

On the production track, scanning relied on an off-the-shelf solution. Walmart had its own Zebra SDK, a C++ machine learning library, but it had to be brought into Android before the app could use it.

What I built

  • An AR view of the aisle. An ARCore prototype for assortment and sorting aisles that recognises products and overlays their information in real time: price, inventory count and model details.
  • Built for pickers and back-office workers. The prototype was made for people working in distribution centres and back offices, on constrained handheld devices, so I optimised the camera pipeline and the AR rendering for them.
  • Zebra SDK integration through JNI. I integrated Walmart’s Zebra SDK into Android via JNI, replacing the off-the-shelf scanning. JNI is the bridge that lets Kotlin code call a native C++ library.
  • A warehouse inventory audit tool. It detects mismatches, produces reports and exports to CSV.

Outcome

The Zebra SDK integration replaced the off-the-shelf scanning and is faster and more accurate than generic ML Kit. The inventory audit tool was delivered with mismatch detection, reporting and CSV export.

The AR work was delivered as a proof of concept and a research asset for Walmart’s AR roadmap. The next step was in-store directions to a product; my contract ended before that began.

All work