Research & Projects
Over the past few years, I've been fortunate to have been involved in research & analysis with Professors Russell Schwartz, Jian Ma, James Langmeed, and Andrew Gow as well as projects at Amazon, IBM, InterSystems, Harvard University, Carnegie Mellon University, and Fondazione Bruno Kessler. Below is a summary of my work.
​
Deep Learning Scavenger Hunt App | Scalable Production Grade Pipeline for Transcript Analysis | Question Suggestion Engine | Senior Thesis | iGEM | EMR Platform
Deep Learning Scavenger Hunt App
Players of this game are given a prompt (e.g. “Capture an Red Car”) and either have to go outside and take a picture of one or draw an image. Once the player has captured an image, they upload it to our web app, which will validate if the child succeeded in capturing the prompt correctly. We hope, our app will encourage children to spend time outdoors or more creatively indoors. Our project utilizes computer vision, using convolutional neural networks (CNN) to identify objects in the image and takes advantage of the BERT transformer architecture to deal with question prompts and responses.
​
Building a Scalable Production Grade Pipeline for Transcript Analysis
Virtual agents (aka chatbots) have become a popular alternative for chat on company websites. Our team support the chat feature for IBM product pages, and wants to understand how interactions with virtual agents differs from interactions with live agents. Additionally we want use these differences to improve the IBM products and the overall customer experience. In order to assess and compare interactions we created a Transcript Analysis Engine. Transcript Analyzer is a scalable production-grade pipeline that uses NLP, ML, and Data Visualization to extract meaningful insight from conversational data.
​
Click here to see the code!
Question Suggestion Engine
Chatbots are an excellent way to engage with users on a webpage. At IBM the conversational guidance provided by live agents is an extremely important part of the customer journey. Since chatbots are rule based machines trained on intents and entities to provide responses, we built a question suggestion engine that could help chatbots guide conversations the way a human representative might. We used intent classification instead of semantic similarity to better identify the current intention of the user and the direction of user conversations. This invention was selected to be filed as a patent in August 2019. I was also selected to present this project at the following conferences: Grace Hopper 2019 and Women in Statistics and Data Science.
​
If you would like to learn more, click here.
Senior Honor's Thesis
​Developing an Accurate Pipeline for Predicting Tumor Progression in Cervical Cancer
Phylogenetic methods applied to fluorescence in situ hybridization (FISH) data are effective at inferring these phylogenies in single tumors by analyzing gain and loss mutations. In this study, I generated a computational pipeline that that takes FISH data for patients with primary and metastatic tumor phenotypes, extracts information related to gain and loss mutations, and incorporates Principal Component Analysis (PCA) and Support Vector Machines (SVM) to generate a linear hyperplane that enhances the classification of the given data. When applied on a set of cervical cancer data, my pipeline able to achieve a 98% accuracy in classification.
If you would like to browse my thesis please click here
Computational Genomics | Term Project
Comparison of Different Clustering Methods on Breast Cancer Tumor Data
This project was address the fundamental over diagnosis problem by finding a better method of determining if a tumor is malignant and likely to be harmful to the host. Using phylogenetic trees constructed from Flourescence in situ Hybridization (FISH) data, we found a set of tree statistics to distinguish malignant metastatic tumor phenotypes from benign ones.
Browse my paper here
International Genetically Engineered Machine (iGEM) Team​
Development and Characterization of Fluorescent and Luminescent Biosensors for Estrogenic Activity | PLOS
​
Leveraged computational methods to build a low-cost, portable biosensor for estrogen detection in water supplies. For this project our team won a Gold Medal & Interlab Award at the 2015 iGEM Jamboree.
​
Fundamentals of Programming and Computer Science | Term Project
Easy-Med an Intuitive EMR Platform
​
Developed an Electronic Medical Records (EMR) user interface that not only stores and displays patient data, but also analyzes quantifiable patient variables and represents them graphically.
If you'd like to browse my code and paper click here