Led a team of four interns in developing a machine learning framework for detecting healthcare fraudAnalyzed datasets of inpatient and outpatient cases to identify prominent features that will determine fraudConducted research in feature engineering and model section, building SVM, random forest, neural network models for effective fraud detectionUtilized AWS S3 and Jira to track individual and group progress on project developmentCo-authored whitepaper detailing final findings and progress throughout the project

Added functionalities to fMRI imaging toolkit using React.js and D3.jsOther functionalities created include matrix generation, Circos plot generation and 3D visualization of brain networksApplication allows for complete visualization of fMRI scan and simplifies timeseries data for userUsed NetworkX to generate Louvain community calculations of timeseries plots of fMRI scans

Collaborated with Dr. Josepheen Cruz to develop a machine learning model to predict the gestational age of a fetal MRI scan from features located in the brainDetects signs of fetal abnormalities that indicate signs of congenital heart disease and provide an estimation for due dateDeveloped models that scored 90%+ accuracy which included SVM, logistic regression, and ImageNet neural networksCombated severe dataset bias using feature engineering techniques, including Pearson’s R correlation and PCAImproved model accuracy from 35-55% to 72% by creating percolated adjacency matrices to represent fMRI fetal scans and used graph convolutional models to predict gestational age