Harsh Kumar

PhD Student · Artificial Intelligence · Human Computer Interaction · University of Toronto · harsh@cs.toronto.edu

I am a 2nd year PhD student working with Dr. Joseph Jay Williams. My primary research focus is on using reinforcement learning algorithms (like contextual bandits) to optimize the interaction of users with technology. As part of Microsoft's AI for Accessibility grant, I have been working on an adaptive text-messaging system to personalize the content and delivery of text messages for the mental well-being of users, based on the contextual information of the users. Another line of work focuses on how pre-trained general AI models could augment user workflows. I aim to bridge my research in adaptive experiments with pre-trained language model research to explore the design space of self-improving and personalized conversational agents to improve the everyday lives of users.

Experience

Graduate Research Assistant

University of Toronto
  • Human-Computer Interaction: Exploring the design space of using Large Language Models (like GPT-3) as human augmentation tools in the context of mental health and education (part of Dynamic Graphics Project lab). Ongoing work includes the use of GPT-3 as a scaffolding tool to improve student learning outcomes through a web-based interface and a well-being coach to help users reflect and improve their moods.
  • Mental Health: Developing an adaptive text messaging service to help people better manage their mental health, in collaboration with Northwestern University and Mental Health America, as part of Microsoft's AI for Accessibility grant. I focus on exploring the role user's contexts play in the reception of these messages and then use reinforcement learning algorithms to improve the effectiveness of interventions in real-time, while incorporating feedback from domain scientists and clinical psychologists.
  • Artificial Intelligence: Developing and applying (contextual) multi-armed bandit algorithms for personalizing interventions related to social good. Some of the questions I explore are how can we balance trade-offs between reward maximization and statistical analysis, and how can we visualize and infer from experimental data to improve the design of the experiment in real-time.
September 2021 - Present

Graduate Teaching Assistant

University of Toronto

CSC428: Human-Computer Interaction, CSC2558: Multidisciplinary HCI, CSC318: Design of Interactive Media, CSC108: Introduction to Computer Programming.

September 2021 - Present

Software Engineer

J.P.Morgan Chase & Co.
  • Worked on a Long Short-term Memory (LSTM) network for Natural Language Understanding (NLU) module which could be improved using users' feedback. The project involved working on an end-to-end system involving the design of a user interface, the building and deployment of scalable models, and databases with continuous integration and delivery (CI/CD) pipelines.
  • Skills: Python - Django, Flask, Tensorflow, PyTorch. Tech stacks - LAMP, MERN. CI/CD.
January 2020 - August 2021

Research Analyst Intern

McKinsey & Company
  • Involved in the development of a semi-autonomous unmanned aerial vehicle (UAV) to optimize warehouse operations. Responsible for developing the vision system for object detection and avoidance using visual simultaneous localization and mapping technology.
  • Skills: Pytorch, OpenCV, NumPy, Microcontrollers.
May 2019 - January 2020

Programming Co-head

Students for the Exploration and Development of Space (SEDS-VIT)

Led the software teams of SEDS-VIT Projects, which took part in a number of international robotics competitions like AUVSI-SUAS, European Rover Challenge etc.

July 2018 - September 2019

Education

University of Toronto

Ph.D. in Computer Science
Courses: Introduction to Machine Learning, Design of Self-improving Systems, Ubiquitous Computing, Computational Social Science
September 2021 - Present

Vellore Institute of Technology

Bachelor of Technology, Computer Science and Engineering
July 2016 - July 2020