• Predicted tuberculosis from chest X-rays using deep learning convolutional neural network (CNN) built from scratch with 84% accuracy and transfer learning with 88% accuracy using Python Pandas, Numpy, Scikit-learn, Matplotlib, and Pytorch on AWS. • Implemented CNN model to detect 14 advanced lung diseases on over 100,000 chest X-rays(15G) released by NIH, resulting in mean AUC 0.75 which outperforms one of the previous state-of-the-art results by 1.2%. Wrote custom cost function to handle data imbalance. Deployed model using Python Flask.

• Conducted ad hoc data analytics, statistical hypothesis testing, and dashboard using R, SAS, SQL, and Tableau to help policy makers deepen understanding of US PK-12 education status, including student engagement, gender gap, education inequality.• Led and coordinated Department of Education's annual data products release. Developed data processing pipeline; automated data review, reporting, and documentation generating processes.

• Measured key metrics to monitor United States PK-12 student academic performance, school and teacher characteristics using NAEP complex multi-stage survey data, with techniques such as variance estimation using Jackknife, bias analysis using Chi-square test, t-test, multiple comparison, and HLM.• Built analytical tool in R enabling researcher with less programming experience to analyze NAEP data more efficiently; including guiding junior staff.• Fitted multiple linear regression with fixed and random effect and factor analysis to predict the difficulty of Grade 4 and 8 reading items with Coh-Metrics factors, improving NAEP reading assessment effectiveness.• Validated NAEP math and history assessment framework by implementing stratifed bootstrap re-sampling method, creating 10,000 hypothetical assessment samples to determine Mathematics framework content coverage variability; validated results with the most recent 5 assessments; presented research study result at AERA annual conference.

• Extracted education indicators from household surveys to gain insight of local school district education quality using STATA, SPSS, and SQL.