I developed a data processing pipeline to convert RNA-Seq data to matched Alpha-Beta TCRs from 30 human patients across 7 years. I also developed a suite of statistical and data science analysis tools to determine how the TCR repertoire changes with aging and across the 7 year gap.
I developed methods and algorithms to model the dynamics of rigid fiber-like particles and analyze their clogging properties around flow corners/bends in microfluidic devices.
I developed an innovative vectorization algorithm for my CFD-FSI simulations to speed up intra-particle interaction computations. Benchmarked against other methods, the vectorization results in an approximately 100x decrease in computation times.
Researched and developed methods for CFD-FSI simulations to model and quantify particle transport dynamics in constricted flow geometries
I developed a predictive framework consisting of CFD-FSI simulations and image-processing techniques to quantify particle bending due to changes in the particle’s mechanical properties.
Used the libroadrunner Python library to implement pharmacokinetic (PK) models and SBML simulations of how the body processes caffeine due to ingestion of caffeine-based food products.
Using a large-scale database of food images (https://data.vision.ee.ethz.ch/cvl/datasets_extra/food-101/), I built a MobileV2Net convolutional neural network (CNN) to classify approximately 5,000 images into 101 distinct food dishes.
Using quantifiable metrics of NBA All-Stars, I built 4 machine learning classification models (logistic regression, support vector, random forest, gradient boost classifier) to predict which players from the 2019 NBA Draft were most likely to become all-stars based on their college statistics.
I developed a neural network to solve partial differential equations that govern fundamental phenomena.