In this project, I devleop a small CNN model that accurately predicts the 15 facial keypoints of an face. The dataset is from a popular Kaggle competition. I preprocess the data using various data augmentation methods, and train a CNN model that uses a LR scheduler. 
In this paper, we developed a casual fairness pipeline for observational data. This pipeline can be applied to analyze classification outcomes and give insight into the effect of statistical fairness mitigation algorithms. 
The goal of this project is to create a system that utilizes a self-evaluation policy to determine whether a task has been completed, and self-identify errors made during task execution without a human evaluator. 
Problems such as identifying Schizophrenia and Alzheimer’s have shown great results when using neural networks. In this paper, we analyze the results of using transfer learning to identify MDD in MRI scans. 
We propose a new architecture based on transfer learning and SNP data to describe the sensitivity of arabidopsis accessions in response to phytohormone perturbation. 
The goal of this project was to reimplement several key aspects of a recent paper by Corso et al. concerning a new method for learning embeddings for biological sequence data, as well as to extend the analysis to novel data and train-test split methods. 
We propose a methodology that uses an ontology—a knowledge base on chemical engineering model forms, relationships, and fundamental laws—in tangent with a novel search algorithm that structurally and contextually matches an equation to its corresponding first-principle definition. 
Analyze and identify hidden/indirect bias that affected the outcome of colorectal procedures in U.S. hospitals. 
In this project, I create an analytics dashboard to track COVID-19 globally and nationally using differenct filters. The dashboard continously gathers data hourly to provide real-time analytics. Analytics can be stratified by race, gender, country, state and many more! 