-Applied computational drug repositioning models to sepsis datasets for the identification of targeted therapeutics in specific sepsis endotypes-Assisted in refinement and redevelopment of sepsis endotypes using unsupervised machine learning techniques on large-scale transcriptomics datasets-Expanded internal datasets by scraping and cleaning transcriptomics datasets and metadata from online data banks including GEO-Integrated external databases from DrugBank and deployed using MySQL for interfacing with transcriptomics pipelines-Prepared manuscript for submission to academic journal and contributed to writing government grant proposals

Predicting Preterm Birth using Electronic Health Records-Investigated structured data from electronic health records (EHR) of UCSF hospitals of >30,000 patient pregnancies with 10,000,000 lab measurements, 2,900,000 drug exposures, and 5,500,000 comorbidities-Utilized unsupervised and supervised machine learning to identify empirical factors associated with preterm birth and to predict the likelihood of premature delivery-Implemented collaborative approach training/testing and validating on data from two separate health systems showcasing similar predictive results across two different populations, with manuscript submitted-Calculated effects drug exposure in early pregnancy on rates of preterm birth in matched cohortsMaternal/Fetal Immune Response in Pregnancy-Analyzed high-throughput sequencing data to characterize maternal and fetal immune repertoire differences between term and preterm pregnancies-Measured B-cell receptor and T-cell receptor diversity, clonal expansion, and preferential usage of variable and joining gene segments, in addition to finding similar clones across samples -Found potential markers of preterm birth by comparing repertoires from term and preterm pregnancies, including evidence of shared maternal/fetal response to infection and notable V gene usage differences Leveraging Transcriptomics for Computational Drug Repurposing -Generated differential gene expression signatures by performing meta-analyses on results from publicly available transcriptomics datasets-Developed computational drug repurposing pipeline to calculate “reversal scores” from publicly-available transcriptomic data to identify existing drugs as therapeutic candidates for the prevention of preterm birth (Le et al., JCI Insight 2020) and for the treatment of COVID-19 (Le et al., Scientific Reports 2021)-Taught 5 workshops for graduate students on applying drug repositioning pipeline to various conditions, including preterm birth, breast cancer, and COVID-19

• Devised and led projects to investigate magnetoelectric response of 2D nanowire arrays to changing external stimuli in extreme conditions• Performed large-scale micro-magnetic simulations and created mathematical framework to model physical phenomena and validate with experimental results • Maintained liquid helium cryostats and designed system to recover evaporated helium, reducing liquid helium costs by 40%• Programmed LabVIEW VIs to automate data collection quickly and efficiently• Organized interdisciplinary collaboration with external research groups to facilitate experiments, construct theoretical frameworks, and interpret experimental results• Published 3 articles in respected physics journals and presented results at research conferences

- Performed data analysis for spin ice studies- Assisted in maintenance and operation of dilution refrigerator and helium-4 cryostats

• Teaching assistant for PHYS 211, a calculus-based introductory physics course (mechanics and kinematics)• Guided 100 students per semester through problem-based discussion sets and lab experiments• Graded problem sets, administered exams, and held weekly office hours.

Grader for various physics courses: Physics 10 (General Physics), Physics 105A (Analytic Mechanics), Physics 105B (Analytic Mechanics)