Built prediction model of stock volume and returns using linear models and gradient boosted trees. Discovered a trading strategy based on predicted returns; back tested this strategy to verify the strategy is profitable.

Build scalable forecast models of promotion sales of Amazon devices across multiple product lines (e.g., fireTV, eReader, Tablet). Designed price experiments for device price elasticities.

Found that online sellers on Alibaba gain 40% more revenue from market trend information provided by the platform, using causal inference techniques (diff-in-diff and synthetic control).- Built a structural econometric model of sellers learning about market trend and subsequently optimizing product assortment decisions.- Implemented MCMC with Gibbs sampling to overcome the computational burden in estimation.- Optimized the information design problem for the platform. Presented results to Alibaba’s leadership; the recommended design was subsequently implemented.

• Analyzed the impact of online sellers’ reputation on consumer purchasing behavior.• Explained economic concepts to the seller reputation teams at the data provider.• Implemented panel data techniques and causal inference.• Built a structural model taking into account the consumer searching behavior and seller growth.

• Designed the course covering a programming tool, with an application to the estimation of multiple econometric models.• Taught the course to 14 students of different backgrounds, supervised development of students’ individual research projects.

TA'ed two courses• Applied Econometrics in Microeconomics (Graduate course): explained various econometric methods (OLS, fixed effect, IV regression, diff-in-diff, regression discontinuity, dynamic panel).• Market Power and Public Policy (Undergraduate course): explained economic theories of consumer demand and firm competition to students of different backgrounds.