Used an ML model, TabNet, that takes in DNA methylation data as input to predict the origin of circulation-free DNA in the bloodstream. This could help physicians use blood samples to gain a deeper understanding of disease and damage in their patient's bodies. I presented my findings to my peers at SURF Seminar Day.Received James G. and Elaine Peterson Fellowship for work in bioinformatics.

Used Google Cloud Platform to code a program that scaped a customer-selected Discord serverfor potentially profitable tradesAutomatically submitted these trades to the customer’s broker account using the Alpaca APIEnsured trades were not made twice by storing previously made trades in a SQL database, and cross-checking against it.Created front-end on Appsheet for customer interaction with the web scraper and linked GCPproject to it using Google Sheet’s AppsScript functionality

Learned SQL, and looked through the lens of a data scientist/engineer by working on a project to analyze the content Netflix has put out over the years. Presented analysis to Facebook data scientists/engineers and other challenge participants

Worked with the Hsiao lab at UCSD to conduct airway segmentation from a CT scan, using a U-Net. The end result was a multi-channel cascaded 2D U-Net model (using Keras), that could detect the trachea and mainstem bronchi (separately) with 95% (Dice similarity coefficient) accuracy. This can be used in a pipeline to output the airway caliber and structure, helping diagnose and assess the severity of respiratory diseases such as COPD.