Andrew Nagata

An Objective-C iOS app that helped golfers record their swing and track improvements by providing a visual diary of progress over time. The goal was to create an app that made swing recording automatic. We developed a custom classifier for recognizing swings to detect swings, edit them down, and save them in the user's journal.

This project was internally incubated and funded. I performed all the product development and engineering while working with contract UI designers to generate the look and feel.

The app uses Firebase for authentication, notifications, cloud storage, and database management. With a paid subscription, users could upload unlimited videos and use the app across multiple iOS devices.

The classifier was created using hundreds of swing videos scraped from social media. The app uses OpenCV to pre-process video frames for classification using Apple's Core ML.

An Apple Watch app was included to offer control of the app from a distance.

An iOS app written in Objective-C that helps golfers putt under pressure. DigiPutt is a putting game that turns an iPhone into a hole by using a combination of real-time image classification and Core Motion. To score well, the user must make several putts in a row from random distances. The app tracks every session and keeps a running "handicap" of the player's ability.

The app uses OpenCV to pre-process video frames and Core ML to determine if the user has attempted a putt. Capturing hundreds of putts to train a classifier presented a challenge. Therefore a separate app and Bluetooth device were created to capture and record processed images of a ball rolling.

This project was internally incubated and funded. I led the development of the entire app from start to finish, including the UI/UX, engineering, machine learning, product design, and publication.

Piñata Party is a multiplayer VR game where players run in place to collect as much candy as possible. Designed to be a light form of exercise, players run in place to compete. It allows up to four players to compete from anywhere.

The game was written in C# with the Unity game engine. Machine learning tools were created in tandem to build custom motion classifiers for locomotion. Combining motion data from the headset and controllers, the game could accurately detect whether a player was at rest, jogging, or running.

This project was internally incubated and funded. I was in charge of all product design, UI/UX, engineering, game design, and publication, and I worked with contractors to perform initial concept art and sound design.