Write an R package to analyze, visualize, and interpret Nanostring data (including calculating DEG, conducting GSVA/GSEA, and performing GO enrichment analysis/KEGG pathway analysis)Write a UI in R Shiny to more easily access/use the R package

Extracted count data from several types of sequencing data using the TopHat-Stringtie-Ballgown pipeline and the GDC pipeline for genomic data analysis (e.g. RNA-seq, DNA-seq)Wrote statistical programs in R to help scientists identify about 10+ relevant genes from pool of about 1000+ genes to liver cancer treatment effectsWrote statistical analyses in R to identify about 100 genes from a pool of 3000 genes that influence lung cancer treatment effectsAnalyzed and interpreted predictive models for principal investigators to understand experiment results.

Research Projects:- Self-driving RC Cars, Jan 2019-May 2019Trained self-driving remote control cars using RL methods including SAC, DDPG, and PPO using Python (pytorch, simulation in Unity).-Mixed Model Thompson Sampling for Multi-Armed Bandit Problems, Mar 2018-May 2018Adapt mixed model Thompson sampling methods to contextual bandit problems with spillover rewards and inter-arm interactions; performance then compared to UCB methods, giving around a 5% improvement in regret.- Human Trafficking Detection with Recurrent Neural Networks, Jan 2018-May 2018Augmented a text-based NN model for detecting human traffickers with a CNN model in Python (pytorch) to detect potential victims of human trafficking (trained on roughly 2k images for roughly 1k ads).- Zillow Price Error Prediction, Aug 2017 - Oct 2017Predicted real estate price errors using a deep neural network regression model in Python (keras with tensorflow backend) based on various home data (including tax data and physical details)- Modeled disease progression in tobacco plants using logistic regression in R.Compared the model for spread of disease in South Carolina to one developed for farms in North Carolina, and suggested potential causes of differences

Forensic Geolocation using Deep Learning ClassificationBuilt a predictive spatial classification deep neural network model to trace the origins of dust samples based on genetic signatures (roughly 10k signatures for about 200 locations). Proposed model achieved an accuracy of roughly 60%.

Fall 2015 semester: Statistics by ExampleSpring 2016 semester: Statistics by ExampleFall 2016 semester: Fundamentals of Linear Models and RegressionSpring 2017 semester: Fundamentals of Statistical Inference IIFall 2018-Fall 2019: Introduction to SAS ProgrammingSpring 2020 semester: Introductory Statistics for Engineers• Tutored students in basic statistical concepts and diagnostic methods for linear regression.• Assisted professors with classroom instruction and record-keeping.• Met with students during weekly office hours.

Developed time series model in Python to forecast monthly demand for inventory control of 100+ food products at 1000+ locations. Created new predictive models of sales volume from sales history in Python and SQL to improve performance from existing model by roughly 2%

Developed and tested fundamental data analysis framework (includes linear models, activation functions, and time series models) in Python for the data science group to solve data problems.Modeled and predicted daily water usage data for commercial, residential and industrial sites in a city using ARIMA time series models in Python as test case for new data analysis framework.

Added color palette for visualizing brain measurements and loading bar features to brain imaging software using Java and MATLAB

• Tutored students in complex analysis and methods of solving differential equations (including Laplace transforms, Fourier transforms, and Green’s functions).• Assisted professors with classroom instruction and record-keeping.• Prepared presentations for lectures and deliver lectures.• Met with students during weekly office hours.• Created and wrote materials such as visual aids, supplementary notes, and sample problems.

• Applied Bayesian statistics to social science studies.• Developed Python programs for analyzing decision and reaction time data for social science studies using Hierarchical Bayesian estimation of the Drift-Diffusion Model.• Presented research results to Caltech students and professors.

• Tutored students in methods of solving differential equations (including Laplace transforms, Fourier transforms, and Green’s functions).• Assisted professors with classroom instruction and record-keeping.• Prepared presentations for lectures and deliver lectures.• Met with students during weekly office hours.• Created and wrote materials such as visual aids, supplementary notes, and sample problems.

• Used MATLAB’s Stateflow and Simulink modules to write programs for experimental trials for measuring an individual’s hand-eye coordination using specific equipment in the lab (KINARM)• Reported results to professor.