I completed my honors thesis in the Department of Computer Science at Dartmouth College with Dr. Soroush Vosoughi, where I utilized natural language processing and deep learning to discover the linguistic features that are important in pain relief using transformer-based models. I have experience manipulating and analyzing large data sets, and I conducted a LIWC analysis to quantify sentiment. I also used large language models, such as GPT-4 and BERT, to generate data through prompt engineering and classify sentences respectively. Furthermore, I analyzed classification model weights using SHAP scoring to understand which features of the model are predictors for pain relief.

I was selected as a TA based on my strong performance in the course. As a TA, I guide students in learning Python by leading weekly recitation sections and office hours. I also grade short assignments, labs, and exams. (Spring 2020, Winter 2021, Fall 2021, Spring 2022, Fall 2022, Winter 2023)

As James O. Freedman Presidential Scholar, I collaborated with the Jacobson lab from September 2021 to December 2023 on a project focused on the creation of a therapeutic chatbot designed to assist users during times of crisis when professional help might not be readily accessible. I assisted in the development of a suicide ideation detection model to complement the chatbot. This undertaking included data mining from a substantial online discussion platform dedicated to mental health topics, and I used this data to develop, train, and test the model. Following this experience, I had the opportunity to contribute a chapter in the new edition of a prominent neurobiology textbook, where I discussed the significance of digital phenotyping and the passive collection of user data from online platforms in research. During this time, I contributed to a publication in the Journal of Affective Disorders.

I worked in Dr. Alexandra Golby’s lab at the Brigham and Women's Hospital as a research intern through the TechFoundation program to develop a low-cost navigational system for neurosurgical procedures. I programmed a system that would manually and automatically generate a dictionary of depth images at different camera positions using the Pyvista library. To more realistically simulate images taken before and during surgery, I added noise through holes and values from a normal distribution. I used techniques in deep learning to train a convolutional neural network to infer a camera position from a depth map.

I was accepted into the Undergraduate Summer Scholars Program (USSP) in the NIH Center for Molecular Studies in Digestive and Liver Diseases at the University of Pennsylvania. As part of the Kaestner Lab, I investigated the role of the histone demethylases UTX and JMJD3 in the development of the perinatal intestinal epithelium.

I assisted in an NLP-based project by tagging datasets and creating a classification system for a CT report text extraction model. I also led corollary research by creating a model for extracting information from CTA reports and auto-extracting Marshall and Rotterdam scores. For this project, I have been named as a first author on an abstract published in the Neurocritical Care Society journal, which I presented at the 19th annual symposium in late October 2021. I have also been named as a co-author in three abstracts (American Neurological Association, American Academy of Neurology) and one publication (JAMA Network Open).

I work for the Elwyn lab through the Dartmouth Institute for Health Policy and Clinical Practice to assess the quality of cancer patient education materials through machine learning techniques and natural language processes.