cv

Bachelor

University of California—San Diego (UCSD)

• Double Major in Data Science & Human Biology
• Overall GPA 3.97
• Core Courses.
  • Calculus, Linear Algebra, Statistical Methods, Data Management, Recommender System and Web Mining, Systems for Scalable Analytics, Data Visualization, Probabilistic Modeling and Machine Learning, Data Analysis and Inference, Programming/Data Structures/Algorithms for Data Science, Deep Learning, Intro to Machine Learning, Statistical NLP, Metabolic Biochemistry, Genetics, Cell Biology, Molecular Biology, Human Physiology and Lab, Biology of Cancer, Immunology, etc.

International Baccalaureate Certificate

Windermere Preparatory High School

• Overall GPA 4.37/4

Research Assistant

Prof.Rose Yu’s Lab at Department of Computer Science and Engineering & Halıcıoğlu Data Science Institute of UCSD

• The project builds a data-driven simulator (http://dx.doi.org/10.1016/j.bspc.2015.04.004) that combines simulation and real-world data that accurately predicts a patient’s Mean Arterial Pressure (MAP) and provides reference for optimal flow control of percutaneous ventricular assisted devices (pVADs)
• Reset blood pressure distribution to simulate original data, perform Sim2Real transfer to train a DANN model (Domain-Adversarial Training of Neural Networks) in PyTorch, attempt to reproduce and verify previous simulation results on both myocardial infarctions (AMICGS) and high risk percutaneous coronary interventions (HRPCI) cohorts
• Continue to investigate the performance of different deep learning models, compare performance of the conditional LMU encoder with DANN model by switching motor speed

Research Assistant

Prof. Enfu Hui’s Lab at Department of Cell and Developmental Biology of UCSD

• Studied how N and O glycosylation of CD80 contribute to the different binding levels between CD80 and CTLA-4, investigated the functional mechanism, and how it might affect the function of CTLA-4 as immune checkpoint and suppressive regulation of T cells
• Designed CD80 mutants with different glycosylation sites with MegAlign software for DNA sequencing alignment and A plasmid Editor (ApE) for designing mutations
• Cultured desired mutants in a variety of cells, employed the protein PD-L1 into the system to detect the glycosylation site affected by PD-L1/CD80 interactio
• Utilized techniques and tools including PCR, spectrophotometer, western blot, ImageJ, protein quantification and purification, cell sorting, etc., applied FRET Assays to identify the results

Undergraduate Tutor for BIBC102 Upper Division Metabolic Biochemistry at UCSD

• Held 50-minute discussion session and 60-minute office hour weekly (or with appointment)
• Took responsibility for homework and exam grading
• Gave a 3-hour review session before the exam
• Received 100% recommendation from both professors and twice from students

Group Project Leader

DS3 society at UCSD

• Led a group of 4 undergraduate students on a global health project
• Researched on WHO data to reveal health disparities and inequalities in medical care by comparing marginalized and non-marginalized groups
• Compared whether and how gender plays a significant factor influencing life expectancy, cancer death rates, and health indicators across continents in different countries and regions, visualized the data with Python to showcase the findings
• Planned to adopt machine learning methods to alternator factors and build predictive models

Shadow Intern

Huadu District People’s Hospital of Guangzhou

• Shadowed doctors in outpatient intakes and inpatient ward rounds of Thyroid Specialization and Gastroenterology & Hepatology Specialization for 8 hours on weekdays
• Observed thyroid endoscopic surgery, endoscopic retrograde cholangiopancreatography (ERCP, a minimally invasive interventional procedure), hepatobiliary & gastrointestinal resection and repair surgeries, as well as minor outpatient surgeries, etc.

CSE 151B Upper Division Deep Learning

UCSD

• Learned how to develop methods for trajectory forecasting in autonomous driving from two perspectives, data and models
• Selected 6 cities for data preprocessing, implemented feature engineering, translational variance, data normalization techniques, and combined all the data for model training
• Experimented with different models (constant velocity, acceleration, linear regression, LSTM, Encoder/Decoder, MLP/RMLP), and identified the recurrent multilayer perceptron (RMLP) as the preferable model

• SUMMA_MAGNA _CUM

• Provost Honor