• Gender Shades

    How well do IBM, Microsoft, and Face++ AI services guess the gender of a face?



  • SELECT Explorations

    Algorithmic Bias : Automated Facial Analysis

    MIT Media Lab

    Advised By: Ethan Zuckerman, Mitch Resnick, Hal Abelson


    Machine learning is increasingly part of daily life. We have learned how to use data sets to teach machines to detect faces and predict patterns. However bias in training data leads to bias in the systems that are created. Algorithmic bias leads to exclusionary experiences and practices. Joy is establishing tools and methods for full spectrum testing of computer vision libraries to mitigate algorithmic bias that leads to poor detection of faces that are not well represented in current training sets.


    Citizen Sensing

    MIT Media Lab

    Advised By: Ethan Zuckerman


    I contributed to the Promise Tracker (PT) platform as a software developer on the Citizen Sensing initiative. The aim of the Citizen Sensing pilot was to 1.) explore one sensor type as basis for understanding the practical considerations of employing sensor data for civic action and 2.) to experiment with visualizing sensor data to inform future work in this space.


    Citizen Sensing Overview

    Broadening Participation in Computer Science Education

    Oxford University

    Advised By: Dr. Rebecca Eynon


    For her MSc Education (Learning & Technology) dissertation, Joy conducted a comprehensive mixed methods case study of the Georgia Institute of Technology's Online Master's of Computer Science. The OMS CS was lauded by President Obama, and was the first fully accredited Computer Science masters program to be offered through a Massive Open Online Course (MOOC) system.

    Transforming Global Health Data Collection

    Georgia Institute of Technology

    In Collaboration With: The Carter Center


    Starting in her junior year at Georgia Tech, Joy served as a technical consultant and lead developer on a novel electronic data collection system for resource-constrained settings The system was aimed at improving efficiency in large-scale household surveys of neglected tropical diseases.


    A Novel Electronic Data Collection System for Large-Scale Surveys of Neglected Tropical Diseases

    Social Robotics

    Georgia Institute of Technology

    Advised By: Dr. Irfan Essa and Dr. Andrea Thomaz


    As an undergraduate in the Computer Vision Lab, Joy developed a turn taking game (Peek-A-Boo) with a robotic agent (Simon) that a human could interact with. Tracking eye gaze patterns was chosen as a starting point to assess social interactions and eventually lead to early diagnosis of autism.

    Health Informatics

    Georgia Institute of Technology

    Advised By: Dr. Elizabeth Mynatt


    As a freshman, Joy worked in the Everyday Computing Lab in collaboration with Google Health. She helped develop a web-based platform to track personal health informatics with mobile/online input, themed android application, and established visualization tools.


    Salud!: An Open Infrastructure for Developing and Deploying Health Self-management Applications