John Keane

Therabot is a research project in the Geisel School of Medicine to build a chatbot to treat anxiety and depressive disorder. The ultimate goal of this chatbot is to lower access barriers for treating mental health. I built the mobile application using Flutter. I also developed a workflow to curate training data, train the chatbot on high performance computing clusters, and deploy the chatbot on AWS.

BetMate is a product built in the DALI Lab, and it is aimed to improve engagement for chess by introducing the element of gambling virtual coins. I architected the backend server with TypeScript, Express, Socket.IO, and MongoDB to broadcast chess matches and manage wagers. I also set up some Python microservices on AWS Lambda, which hosted machine learning models to determine betting odds.

whiteboard is a productivity application built in the DALI Lab, and it's goal is to provide many productivity features such as project planning, to-do lists, calendars, and more. The mobile application aimed to be a supplement to the web platform. I developed an interactive calendar with JavaScript and React Native with views for multiple timeframes as well as drag-and-drop event modification.

Boboddy is an acronym generator I built during my free time. The generation algorithm is derived from the Viterbi algorithm, which takes in a sequence of observations and returns the most probable sequence of "hidden" states. The text processing and probability models were developed using Python and NLTK, and the app was deployed using ASP.NET to run the algorithm and Angular to display the UI.