My interests are definitely varied, and the projects I've worked on range from initiatives for diversity in tech to machine learning research to artwork. What follows is a sampling of things I've done over the past few years.

  • Intro to Deep Learning
    Organized a for-credit course taught at MIT for the first time in January 2017. An extensive introduction to the field of deep learning, covering applications to machine translation, image recognition, game playing, image generation and more. Includes hands-on labs in TensorFlow and peer brainstorming sessions. Lecture video here; slides here.

  • Pre-Trial Fairness
    An RShiny app and website to understand the tradeoffs between different frameworks of fairness, with the goal of making fair, data-driven models easier to understand and adopt. The winning team of the New England Machine Learning Hackathon: Hacking Bias in ML hackathon hosted by Microsoft New England.

  • Commonwealth Stats
    An application for non-profit organizations to access and analyze important demographic data from the US Census Bureau in an easy and visual way, for use in grants, reports, or research. Made for the Massachusetts Nonprofit Network with Maryam Archie.

  • Chic++
    A platform celebrating the intersection of femininity and tech, with posts including product reviews and engineer spotlights. We attempt to address the common (perhaps subconscious) viewpoint that it's okay to be a woman in tech as long as you act like one of the bros. Chic++ is a highly visual site targeted at young people who don't fit the "brogrammer" stereotype. A collaboration with Kamilla Tekiela.

  • Horizon
    A chrome extension to provide visually-impaired users with image alt-text for screen readers, using convolutional neural networks to caption images.

  • Personalized Health Insights with Jawbone
    Used data from Jawbone UP bands to build models of how different behaviors affect users' heart rates and workout patterns. Shipped personalized insights around auto travel detection, workout patterns, and heart rate trends. Insights received an App Content Score of double the current average within first 48 hours of release. Published two visual data stories on the official Jawbone blog (first post; second post).

  • Classification of Cultural Districts using Yelp Reviews
    Designed and implemented an unsupervised learning algorithm that leveraged LDA topic modeling, Gaussian Mixture Models, and k-nearest neighbor classification to detect and label cultural city districts. Submitted as part of the Yelp Dataset Challenge.

  • Flipper
    Making the flipped classroom more efficient. An easy-to-use and accesible way for teachers to evaluate student understanding before class by uploading and organizing resources and multiple choice questions. A group project for 6.170.

  • Orientation-Specific Attachment of Polymeric Microtubes on Cell Surfaces
    Created heterogenous magnetic-nanoparticle and polymer microtubes using layer-by-layer fabrication. Quantified orientation-specific attachment of polymeric microtubes on cell surfaces using fluorescent microscopy and imaging. Co-authored paper on results in Advanced Materials.

  • MIT Class Ring Design
    As Vice Chair of the MIT Class of 2016 Ring Committee, led the design of the iconic class ring, and planned the class-wide delivery event.

  • Wall Mural
    Painted my wall with mountain ranges.

  • Neuroscience Notebook
    Graphic notetaking for 9.01, Intro to Neuroscience.

  • Biochemistry Notebook
    Comprehensive notes, diagrams, and review sheets for 7.05, Biochemistry.

  • Graphic Design for GlobeMed at MIT
    Advocacy, informational, and marketing designs for GlobeMed at MIT as the Marketing, Design and Media director.

© 2017 Harini Suresh. All rights reserved.