Pavitthra Pandurangan

Ideated and built prototypes of several GitHub bots that were aimed at improving diversity and newcomer retention in open source projects. Based ideas on related research and successful interventions in other communities that struggle with diversity, such as Wikipedia and Stack Overflow. Prototyped first interaction bot, which comments the first time a user make a post in GitHub issues, a bot that creates a monthly issue to collect newcomer feedback & provide mentorship, and a bot that posts various metrics about a given repository (such as response time, fraction of unanswered issues, etc.).

Co-wrote this paper, which was accepted by the journal Computer-Supported Cooperative Work (CSCW), with an offer to present findings at CSCW 2021. The paper analyzed the introduction, discussion around, and use of the code of conduct in open source communities. Our work sheds light on how maintainers experience the tension between merit-based OSS culture and the growing emphasis and call for inclusivity as open source increases in popularity and commercial use.

This project was made at Tartanhacks 2020. Iris is a web application for those who are visually impaired. Our tool allows users to upload an image of anything they are uncertain about and get more clarity. Answers are crowdsourced from other members of the Iris community and members can upvote/downvote answers. In addition, we use computer vision to give users an immediate classification, independent of the crowd-sourced answer.

This project won the Capital One Best Financial Hack at Tartanhacks 2019. Scam Stopper analyzes is a prototype for an application that would help stop phone call scams by analyzing calls from unknown numbers. We turned speech into text, used APIs to analyze the text, and determined whether or not to flag the call as spam. We alert the user to a potential scam if our thresholds were triggered. We used Flask as our web framework, and used jQuery in our web pages.

I made Galaxy Glider for the my term project for the course 15-112. It is a 3D visualization of the universe evolving. Users are able to explore planets in different galaxies in the application. For the planet exploration mode, I developed an algorithm to create randomized trees in Blender. Watch my term project video!

My team of four won the BNY Mellon Sponsor Prize for Best Financial Hack at TartanHacks for Moola. Moola is a user-friendly and all-encompassing financial application that combines a portfolio analyzer with a day-to-day transactions component. I worked on the backend of this project. I used the webhose API to get current articles about a specific company, and used the IBM Watson's Tone Analyzer API and the Microsoft Text Analytics Sentiment API to provide users with current insights into public/investor opinion on the companies in their portfolio. Our code is on github.

At the J.P. Morgan Code for Good hackathon, my team and I created a web application that connects individuals affected by breast cancer. We won runner-up at the hackathon. After the hackathon, the J.P. Morgan Force for Good picked up our project and developed it for Susan G. Komen Foundation.

For Hack112, my team and I created an engaging educational game to teach kids langauge skills kids using OpenCV. We won the Best OpenCV Hack award. The application has three mini games to help kids practice spelling, learn words, and practice reading. One of the mini games lets you practice tracing letters using a yellow object on a screen that displays the text and a picture representation of the word. You can read more about this project here.

I made a custom animation with keyframes, and grew adept at using sass to get my website to look how I wanted.