Python, SQL, R, Hive, Presto, Git, Machine Learning

Zekai (Zach) Zhu

773-***-**** *******@*.************.*** https://www.linkedin.com/in/zazhu/ https://github.com/zekaizhu EDUCATION

Northwestern University, Evanston, IL, United States September 2019 - December 2020 Master of Science in Analytics GPA: 3.85/4.0

• Relevant Coursework: Analytics Value Chain, Big Data Analytics (Hadoop & Spark), Databases & Information Retrieval, Data Mining, Data Management for Business Intelligence, Deep Learning, Data Visualization, Predictive Analytics, Text Analytics Soochow University, Taipei, Taiwan September 2015 - June 2019 B.B.A in Financial Engineering and Actuarial Mathematics GPA: 4.0/4.0

• Dean’s List for all semesters attended and graduated with honor

• Society of Actuaries (SOA): Exam: Financial Mathematics, Probability; VEE: Corporate Finance, Applied Statistics TECHNICAL SKILLS

• Programming languages: Python (Pandas, NumPy, scikit-learn, Keras), SQL, R, Scala, Shell scripting, Java, D3.js, JavaScript, HTML

• Framework & tools: Git, Bash, Hive, Presto, AWS (S3, EC2, RDS), Tableau, Spark (PySpark, SparkML), Hadoop, MapReduce, SPSS, Azure, Docker, Jira, Confluence, A/B Testing, Predictive Analytics, Funnel Analysis PROFESSIONAL EXPERIENCE

Principal Financial Group Chicago, Illinois

Data Scientist September 2020 – Present

• Optimized investments portfolio by adding new objective functions and constraints to minimize tracking error between asset and liability

• Embedded the optimization pipeline in an interactive website for user to obtain the optimized investment portfolio in Python (Dash and Flask) LinkedIn Sunnyvale, California

Data Scientist Intern – Growth Amplify Team June 2020 – September 2020

• Leveraged Presto, Hive, Python, and Spark to analyze massive amounts of data of app activation, notifications, sessions, and members, etc.

• Employed machine learning classification model to understand the relationship between 110 million infrequent members’ email click and member life cycle upshift; proposed most valuable emails to be prioritized for the next two quarters of email edge building strategy

• Conducted deep dive analysis of onboarding flow to improve email confirmation that will largely lift long-term weekly active users; identified emails with deep link issues that mostly impact members’ mobile experience and persuaded engineers to fix the problem

• Performed funnel analysis of phone address book imports to find massive opportunities to optimize recommendation for connection and help infrequent members connect with frequent members; worked with product and engineering teams to drive corresponding experiments

• Migrated metric datasets by running flows on Azkaban to cope with changing data schemas and designed metrics reporting dashboards NASA Jet Propulsion Laboratory (Learn More) Pasadena, California Data Science Researcher October 2019 - June 2020

• Conducted exploratory data analysis and 3D visualization using Python (Plotly, Seaborn) on massive electron number density data collected from THEMIS probes; detected repetitive cycles and anomalies by applying data mining, machine learning and clustering techniques

(Anomaly detection algorithms, t-SNE, PCA, K-means) to help increase accuracy for future disturbance prediction in the physical system

• Built an interactive web dashboard to visualize the clusters and anomaly detections by leveraging D3.js and html (Learn More) TransUnion Chicago, Illinois

Analytics Consultant - Data Science & Analytics team September 2019 - June 2020

• Employed XGBoost to develop credit risk models using 1500 attributes of credit risk reports to measure potential lift in the prediction of account delinquency; elevated the AUC score of the benchmark model by 2% with the addition of macroeconomic predictors

• Performed mapping visualization of quarterly data in R (ggplot): number of new accounts opened and bad rate for each county and MSA

• Leveraged multivariate time series model VAR and LSTM to predict macroeconomic trends using key consumer credit attributes; employed forward chaining to validate the models and achieved less than 0.5 RMSE in the prediction of fraud-related macroeconomic indices DBS Bank (China) Ltd. Shanghai, China

UNI.CORN Management Intern - Corporate Banking July 2018 - August 2018

• Conducted exploratory data analysis with Python and data visualization with Tableau of customers’ data retrieved from DBS public account and gained insights to determine high net worth clients’ preferences regarding financial articles, business tweets, and financial products PROJECT EXPERIENCE

House Sale Price Prediction Application (Learn More) April 2020 – June 2020

• Leveraged AWS (S3 and RDS), docker and Python to build an automated regression model pipeline to predict New York house sale price

• Implemented offline model pipeline in an interactive web-based application where users get forecasted sale price after providing information Deep Learning Face Mask Detection (Learn More) April 2020 – June 2020

• Implemented VGG16 Neural Network to detect human faces on public images (WIDER Face, MAFA) and classified images into without wearing masks, wearing normal masks, and wearing N-95; achieved 82% validation accuracy Society of Actuaries Student Research Case Study Challenge April 2018

• Led a team of four to analyze household historical data with R and constructed reduction factors model and Markov multiple-state model to predict the future balance of a public long-term care insurance program; conducted sustainability assessment through sensitivity analysis

• Optimized the public long-term care program to yield a 138% increase of its balance in 2028 by implementing solutions such as introducing progressive tax rate with surcharges for childless families, allocating budget for healthy habits advocacy, etc.

Contact this candidate