Huili (Stefanie) Si
adbnys@r.postjobfree.com
Santa Clara, CA, 95051
www.linkedin.com/in/huilisi
EXPERTISE & CERTIFICATE:
Programming: Python, R, SAS (Certified Statistical Business Analyst)
Database: MySQL, SQL Server, Postgres SQL, MongoDB
Machine Learning: Logistic Regression, Random Forest, Support Vector Machine, CNN
Experimental & Statistical Analysis: A/B Testing, Multivariate Testing, Cohort Analysis
Data Analysis & Visualization: Tableau (Tableau Desktop Specialist), Power BI, Excel VBA
Professional Finance Experience: CFA (Level III), FRM (Level II), Time Series
Other Software & Tools: Bloomberg, JIRA, SAP, Cognos
EDUCATION:
Stevens Institute of Technology Hoboken, NJ, USA Aug 2016 – May 2018
Master of Science in Business Intelligence & Analytics GPA: 3.9/4.0(Top 1%)
Huazhong University of Science and Technology Wuchang Branch, China Sep 2011 – May 2015
Bachelor of Finance GPA: 3.7/4.0(Top 2%)
Experience:
Ambac Financial Groups, NYC, NY
Data Analyst/Modeler Mar 2019 – Jan 2020
Performed statistical analysis of $23 Billion portfolios and involved the construction of SQL Server databases to detect abnormalities over 1,000 single risks in Excel VBA, maintaining its risk rating in the stable outlook
Engaged in building Residential Mortgage Loan Credit Model, extending a vanilla logistic regression algorithm into XGBOOST with adding regularization parameters to predict defaults and prepayments in R, adjusting premiums of products and increasing revenue to 120% on a year-on-year basis
Designed customized modules in “Shiny” package in R to create GUI for operating ETL process
Ran and validated financial models quarterly via adjusting assumptions and parameters, discovered the insights and developing automatic interactive reports and dashboards in Tableau (Desktop, Server), presented on quarterly board meetings
Spinnaker Analytics, Boston, MA
Data Scientist Jul 2018 – Mar 2019
Built up a web crawler to scrape unstructured text data, and created Stored Procedures, User Defined Functions, View and Trigger using Transaction SQL, gathered data from multiple channels and provided an accurate representation of actionable data in the hands of business and data analysis
Identified potential monetization opportunities through 4-week A/B testing on the website of gaming company to improve the buyer conversion by 2x through drawing Jackknife confidence interval and calculating statistical significance in Python, making recommendations to incentivize users to purchase gift cards
Explored, caught and delivered business insights and opportunities in conducting Waterfall Analysis and SWOT Analysis during daily status calls
Dimex LLC, South Hackensack, NJ
Data Analyst Intern Jan 2018 – May 2018
Defined customers’ demands as one major metric for the e-commerce platform to customer shopping patterns and provided purchasing opportunities from over 5,000 products to evaluate estimated customer lifetime values in MySQL
Effectively extracted, transformed and loaded live daily data and sales activities between the warehouse and e-commerce platforms (Amazon, eBay), identified customer segments using clustering algorithms on clients in Python, and provided predictive analysis to display market trends to avoid stock-outs, decreasing churn rate from 35% to 23%
Presented funnel charts, displayed factors that influence customers’ behaviors via drill-down graphs and geographic maps in Tableau and automatically generated clients reporting to track customers’ preferences twice a week
Kuisheng Sanjie Microfinance Co.Ltd, Wuhan, China
A financial institution principally engaged in the business of providing finance to the public
Data Analyst Nov 2014 – Aug 2016
Comprehensively evaluated more than 30,000 loan applications, with exploratory data analysis (EDA) in Excel(VLOOKUP, Pivot)to examine data structure and distributions, conducted features engineering and detected abnormal behaviors based on the main features (e.g., annual income, locations)
Fitted 5 classification models on training data of loan applications, followed by cross-validations and grid search to select the optimized parameters, and finally set the thresholds of acceptance to classify applicants, thus enhancing 1.5x accuracy lift of identifying creditworthiness of clients