Guansong Wang

KiwiDrop ^[1] Guangzhou (Remote)

Data Engineer | Data Analyst 2023-01 ~ 2023-12

• Optimized data dashboards that stakeholders relied on; built a frugal ETL process and data warehouse; standardized metrics definitions; set up Looker BI system.
• Pricing model analysis, inventory data analysis, competitors analysis, and financial income statement analysis.

Starlinke ^[2] Shenzhen, China

Data Center Director | Senior Data Analyst 2020-08 ~ 2022-02

• Building data infrastructure.
  Constructed big data warehouse and BI system from the ground up, integrated all operational systems, created Wiki of business knowledge, clarified definitions of metrics and metrics system, and designed behavior tracking data schema.
• Collaboration with business partners.
  Led routine and topical data analysis reports, such as user behavior portrait, inventory analysis, and trend analysis based on product clustering. Participated in and designed data-based applications, such as product placement and recommendation, customer labeling system, automated financial accounting, and intelligence system.
• Management.
  Organized OKR and reviews, weekly training and workshops, and monthly one-on-one communications. Cultivated collaborative and productive environment. Promoted accumulation of procedures and documentation.

Patozon ^[3] Shenzhen, China

Manager of Owned-Shop Department 2019-07 ~ 2020-07

• Transformed manual procedures into systematic tools; discovered potential best-selling products via data mining; monthly sales increased seven-fold in 6 months.
• The department became an independent company. Involved in overall management as one of the founders. Successfully achieved the sale objective and received appending investment. The company was then acquired by Starlinke.

Senior Data Analyst 2019-03 ~ 2020-04

• Developed new features and tools to increase productivity, such as Facebook ads ROAS prediction model, Amazon's new trending product dashboards, Amazon ads keywords combination tools, accounting data integration tools, etc.

Yooli ^[4] Beijing, China

Risk management director 2016/10 - 2018/04

• Designed, developed, and upgraded risk management models and system solutions. Achieved a fully automatic crediting system based on GBDT and the lowest default rate in the industry.
• Led a team of 10 colleagues, coordinated feature engineering projects, and organized weekly group meetings and training workshops.

Data analyst (marketing and operation) 2016/05 - 2018/04

• User classification model (GBDT), the prediction model of overall investment and withdrawal, promotion campaign A/B testing, conversion rates alerting system, and abnormal advertising pattern detection.

Ping An Insurance (Group) Company of China Shenyang, Liaoning

Liaoning district, intern 2015/10 - 2015/11

The University of North Carolina at Chapel Hill Chapel Hill, U.S.

Financial Econometrics, M.S. (Ph.D. ABD)^[5] 2007-2014

• Thesis: Realized Kernels with Moving Average Noise and Optimal Weights.
• Research interest: financial econometrics, asset pricing, high frequency volatility models.
• Teaching assistant: advanced econometrics, time series, finance.

Peking university Beijing, China

Yuanpei Program (College), Bachelor of Economics 2003 - 2007

• Thesis: Microeconomic Analysis on the Quota of Microcredit

Frugal ETL and Small Data Warehouse KiwiDrop

• A Python implemented ETL process, and a small data warehouse based on SQLite.
• It enabled data aggregation across scattered business processes, improved the timeliness and consistency of metrics, and laid the ground work of building the data warehouse.

Big Data Warehouse Starlinke

• Constructed data warehouse based on MaxCompute and DataWorks of Aliyun. Integrated business data from all related systems. Greatly improved the usability of data and efficiency of developing data applications.
• Developed BI system, user labeling system, accounting toolbox, product placement, and recommendation based on the data warehouse.

Intelligence System Starlinke

• Collected open data from e-commerce platforms, online stores, and social media, to follow the latest market trends and to support product development.
• Utilized image clustering and classification models to extract popular categories, colors, and elements.

Automatic Crediting System Yooli

• Designed credit system workflow and implementation plan with the developer team. Managed crediting rules, completed development documents, and implemented some of the modules.
• Trained and evaluated credit score models (logistic regression and GBDT); oversaw the deployment of the models; constructed different models according to various groups of users; accelerated model upgrades from monthly to weekly.
• Achieved fully automatic crediting, continuously improved the performance of the system, and achieved the lowest default rate among competitors.

R Packages Yooli

• Developed and maintained a team R package to simplify and standardize workflows, including database interaction with SQL and MongoDB, commonly used ETL methods, metrics computation, etc.

Volatility analysis based on high frequency records of DJIA component stocks UNC

• Extracted and cleaned intra-daily records of 30 stocks from 1992 to 2013 (TAQ database, 40 billions records).
• Filtered out outliers and construct price processes with different sampling frequencies for each stock and DJIA. Built a database and prediction model of various daily volatility estimators for each stock and DJIA.

Programming

Python, R, Matlab, Shell, C/C++
SQL, MongoDB, Hive, MaxCompute

Language

Chinese (native), English (fluent).