CV

Click here for my latest resume.

Experience

• Snap - Senior Engineering Manager [Oct 2019 - Now]
  • I am the single-threaded owner of a business critical area “Friending” at Snap and led teams of client, backend, and ML engineers and managers across multiple geo locations. We build products, such as friend recommendation systems, nearby friends, friending notification/sms, spam prevention, that help users make virtuous friends on Snapchat and experience our value proposition. Innovations such as EBR, GNN boosted topline Friending OKRs by 2-3x.
• University of Washington - Adjunct Professor [Sep 2018 - Feb 2024]
  • I have a passion for teaching and helping more people into the field of machine learning and data science. I currently teach several graduate-level courses in the Foster School of Business, including Advanced Machine Learning, Deep Learning, and Natural Language Processing.
• Amazon - Senior Applied Scientist and Tech Lead [Oct 2017 - Oct 2019]
  • Forecasting I built ML models to forecast products’ sales velocity in realtime to source high-quality deals and optimize the deal scheduling for Amazon Prime Day 2018 & 2019.
  • Recommendation I led the team in building the recommendation system to personalize and rank content on Amazon Seller Central Homepage and other push delivery channels (emails, push notifications and SMS) in order to maximize user engagement.
  • Downstream Impact I built Amazon’s first seller downstream impact causal model to estimate the incremental long-term economic impact of a single seller-initiated action. Outcomes are used in budget allocation, content recommendation and A/B test reporting.
• eBay - Applied Science Tech Lead [Dec 2013 - Oct 2017]
  • Marketing I led a team of 8 applied researchers to power eBay’s paid internet marketing by improving the bidding strategies on Google and Facebook to maximize ROI. Our models improved the ROI by 27% in 2016, leading to millions more GMB annually.
  • Trust I built a large-scale machine learning model to predict seller risk and reduce the number of defective transactions on eBay via search ranking demotion, which translates to 100+ million GMB lift annually.
  • Search I implemented a topic model based approach to retrieve diverse items based on user buying intents and improved the user satisfaction by 6+%.
• Oregon State University - PhD Researcher [Sep 2006 - Dec 2013]
  • I developed probabilistic graphical models to predict species distribution with citizen science data, which significantly improved the accuracy of species distribution.
  • I developed a generative mixture model in the eBird human/computer learning network to quantify the skill level of citizen scientists in the eBird project.

Education

• Ph.D. in Computer Science, Oregon State University [2006 - 2013]
  • Thesis: Machine Learning For Improving The Quality of Citizen Science Data.
• B.S. in Computer Science, Wuhan University, China [2002 - 2006]

Publications

I have published 20+ papers in ML journals and conferences. Please Publication Page for paper details.