2. PROJECT EXPERIENCE
Unsupervised Learning â Identify Customer Segment with PCA and k-means clustering [link] Jun 2019
âą Identified characteristics of the core customer base for a mail-order company from Germany using Germany
demographics data subset (~890k records) and the companyâs customer subset (~191k records) with 85 features
âą Performed data inspection, data cleaning, data transformation using Python seaborn, Scikitlearn StandardScaler,
Imputer and PCA to reduce dimensions to 25 while maintaining 80% of data variability
âą Utilized k-means clustering from Scikitlearn to identify overrepresented and underrepresented segments of
customers for improved marketing strategy
Deep Learning - Image classification of flower dataset using Pytorch [link] May 2019
âą Built a neural network model on the training set of ~6.5k images of 102 flower species using 3 Torchvision
pretrained models: VGG16, VGG19 and DenseNet121, achieved accuracy of 89% on the test set of ~800 records
âą Created a command line application with modular functions to train and predict any other labeled image
datasets
Supervised Learning - Income classification with Python Scikitlearn [link] Mar 2019
âą Predicted income of individuals who make > 50k from 45k Census Dataset of 45k records using 10 different
classification models such as Ensemble (AdaBoosting, Gradient Boosting, Random Forest), Decision Tree,
Gaussian NB, etc.
âą Tuning hyperparameters with GridSearchCV and achieved 87% accuracy and 75% f1 score on tuned model
Data Visualization with Tableau [link] Apr 2017
âą Created various dashboards and storyboarding with different datasets using advanced Tableau calculations, data
blending, grouping, clustering and Analytics tool
Data Warehouse Design for E-commerce Business Apr 2017
âą Designed data warehouse with star schema using Kimbal Methodology for 5 departments of a beauty
Ecommerce business using Erwin
Supervised Learning â Customer Book Purchase Prediction purchase dataset with SAS 9.4 Mar 2017
âą Predicted the number of customers purchasing from Barnes & Noble in 2007 with 33000 records and 8
demographic attributes using 3 models: Negative Binomial Distribution, Poisson Regression and NBD Regression
with Proc NLMixed, Proc SQL and Proc MI in SAS
Exploratory analysis of Prosper Loan dataset with R [link] Oct 2015
âą Performed Exploratory Data Analysis using R ggplot using the Prosper Loan dataset
âą Built linear regression model to predict Borrower Rate for new loans with R-square of 89%
Supply Chain Distribution Network Design with Excel Solver [Supply Chain Case Competition]
âą Proposed the optimal supply network design to reduce operating expense of a mattress company by 21% while
improving service level to 94%. Tool used: Premium Excel Solver
âą Presented the findings to a board of judges from Walmart, Texas Instrument and APICS
âą Won the 1st place against teams from University of Texas at Arlington and Southern Methodist University
EDUCATION
The University of Texas at Dallas - M.S., Supply Chain Management - Summa cum Laude Aug 2017
âą Honors Deanâs Excellence Scholarship (GPA 3.97)
âą Third-place in Operation Management Competition over 30 teams from 4 schools in Dallas/Fort Worth
âą First-place in 2 Supply Chain Case Competitions over 24 teams from UT Dallas, UT Arlington and SMU
Foreign Trade University (Vietnam) - B.S., Business Administration May 2013
âą Honors American Chamber of Commerce Scholarship