Weiyan Sun

Machine Learning Engineer in Ads Revenue Health.

Research Data Scientist within Video Infrastructure and Developer Efficiency by implementing and scaling modeling and statistical approaches to address key technical and business focus areas. Working with large interdisciplinary team including Data Engineers, Software Engineers, and other Research Data Scientists and delivering impact via data-driven approaches. Tech Leading and onboarding team members.

• Head model developer in charge of implementing and monitoring multiple material models including pricing and credit risk

• Lead DS in the development of building a real-time multi-armed bandit API that has been widely used on direct mail, email, website campaigns across multiple teams and delivered on avg 10% application rate lift and $24MM increase in annualized NPV

• Conducted Sequential Pattern Mining on customer’s journey to identify frequent series of events that lead to dissatisfaction and further fed patterns into XGBoost to rank order customer dissatisfaction level, which achieved AUROC of 0.82 (Patent Pending) • Productionized model above along with a LSTM-RNN challenger model with end-to-end model execution and unit test, pylint test covered; Model scoring output is leveraged by PMs, BAs, agents to identify key opportunities on customer experience

• Investigated income inequality in 8 countries in Asia and South America during 1980-2010 using microdata from IPUMS-International and integrated the shapefile with geographic information to implement state-wise analysis • Performed multiple data mining processes using R to evaluate three standard explanations of inequality

• Conducted Exploratory Data Analysis and designed a multi-layer machine learning algorithm with data cleaning and feature engineering on 80 million transactional records to assess potential fraud activity • Trained multiple models using Logistic Regression and Gradient Boosting by dividing the features into three groups (consumer end, processing end, interaction) and applied ensemble method, which achieved accuracy of 0.85 on 5-fold cross validation • Proposed a recommendation system for fraud score with optimized cost-benefit analysis and presented it to client

• Conducted Benchmarking between Audi’s LED Matrix Technology and Mercedes-Benz’s Multi Beam Technology and between Audi OLED Swarm System and BMW OLED System • Organized a customer survey and developed a market analysis during the Chinese Grand Prix of Formula One and Motor Show Exhibition • Identified the market reactions on the new Audi light prototypes in China and provided improvements and ideas based on customers’ opinions • Led a team of 23 to design new lights prototypes by adding more safety and emotion elements using 3D printing and presented the report to the Audi project manager

• Teamed with EW Coordinators to settle breakpoint of components in SAP system properly and analyze the components’ demand for Part Trial Run • Improved a mathematical model based on lowest warehousing costs, which achieved 5% reduction in warehousing cost • Assisted in accomplishing a DFSS (design for six sigma) project about Seal System Optimization for Closing Energy of SGM285 Front Side Door and contributed to an 80-page report • Selected Anti-Lock Braking System module supplier by performing several Technical Assessments