Jeffrey He

• Researched, designed, and implemented a new workflow to improve the monthly support case review process (500 cases/month) for the Azure Stack team
• Created Python algorithm that predicts action items for certain cases

Explorer Intern

• Built a new Microsoft Store feature (using Angular, Typescript, C#, and Jasmine) that provides publishers a detailed history of their product submissions broken down by category
• Feature shipped to all Microsoft Store publishers

Teaching Fellow

• Taught Machine Learning (CS181), Economics and Computation (CS136), Artificial Intelligence (CS182), Functional Programming (CS51), Intro to Computer Science (CS50)
• Led weekly sections of 20 undergraduates, held office hours, designed and graded problem sets

Undergraduate Researcher

• Ran research study on 600 subjects regarding the interpretability of ML algorithms
• Published paper at AAAI HCOMP ’19 (“Human Evaluation of Models Built for Interpretability”)

Harvard University

• Harvard Open Data Project (President)
• Harvard Computer Society (Director of Education)
• Teaching Fellow (CS50, CS51, CS182, CS136, CS181)
• Harvard Data to Actionable Knowledge Lab (Researcher)
• Student Mental Health Liaisons, Intramurals Rep (Mather House)

HODP is Harvard's first open data catalog. As a group, our mission is to use public Harvard data to increase transparency and tackle issues on campus. We’ve used data science and data visualization techniques to analyze data on student government election turnout, faculty gender disparity, and trends in concentration popularity.

This Python script allows you to analyze your FB messages. Specifically, for a given conversation, it prints out a map of the messages you (and others) have sent over time, the average word count of your mesages, and is set up in a way that lets you add any new features easily. I used this for a bootcamp as the Director of Education of the Harvard Computer Society.

Datamatch is a matchmaking service used by ~4000 students each year at Harvard built with Ruby on Rails and React. As the web lead, I led the team in designing the credits and archives page.

The ProPublica scraper is a python script that uses the ProPublica API to retrieve data in CSV format on any list of non-profit US organizations. We used the script to extract data on non-profit student organizations at Harvard.