I’m a computer science student at Northeastern University with a math minor, on co-op as a software engineer at Chewy in Boston. I work on the spots where information gets lost as it moves between people and systems, and nobody notices until something breaks.
Three questions turned into projects. How does a team find the disagreement its Slack thread never spelled out? Who pays for an AI request when the stream cuts off before it reports its final usage? How do you find out what you don’t understand before an exam does it for you?
Those questions became Ringi, Phantom, and Kizuki: a Slack bot, an AI gateway, and a study tool that runs on your own computer. Each one keeps the evidence behind what it tells you, so you can check it later: who agreed to a decision, what a request cost, which sentence of your notes a question came from.
Working across the whole system
At Chewy I work on labor and capacity planning: the Snowflake pipelines that build the forecast, the AWS setup they run on, and the tools that get the numbers to fulfillment teams. In open source, I fixed a Prisma bug where the SQLite driver adapter dropped the milliseconds from DateTime values, added tests for it, and corrected a page of the docs.
I like working across those layers because each one shows you something the others hide. The interface shows the assumptions the database schema never wrote down. A production failure tells you more about the data model than the data model does. Sometimes a change to how people work removes a problem that looked like it needed code.
Smaller projects
LeagueIQ reads a League of Legends match through its decisions, like objective fights, rotations, and who joined which fight, because the scoreboard shows who won and the timeline shows why. For a music course, I used Python to draw Le1f’s “Wut” as a listening score that shows its rhythm, texture, and repeated parts in one picture. Both projects take a finished record and find a way to see its structure.