IG Shopping Relevance

* Related Ads: I initiated and lead a cross-app (FB/IG/Messenger) workstream called related ads, which aims at realtime intentional ads delivery. 2% company wise ads score (internal indicator of long term revenue) win.

* Format Machine Learning: one of the early members personalizing what format to render given a fixed <user, ad> pair.

* Multi Task Learning: I initiated and lead the multi-task learning, model ensemble, and task dependency modeling efforts for major Meta's ads ranking models. 1% company wise ads score (internal indicator of long term revenue) win.

PhD student working on Information Retrieval, supervised by Prof. Luo Si. My research broadly covers recommender system and web search. Specifically, I've done topics such as matrix factorization, collaborative filtering, hashing, learning to rank, search session mining, context modeling, structured svm, query reformulation, query expansion, cost aware feature selection, mobile app recommendation, sponsorship/advertising, etc.

Research Intern at iDST, Seattle, supervised by Prof. Luo Si and Dr. Jun Wang. My work is to design sponsored mobile app recommendation algorithms that maximize the sponsorship profit for appstore meanwhile maintaining the recommendation quality. We achieved significantly better quality-profit balance than the existing recommender system on Alibaba's YunOS appstore (http://apps.yunos.com/).

Research intern jointly sponsored by Microsoft Research, Redmond and Bing, under the supervision of Dr. Jianfeng Gao and Dr. Guihong Cao. My work at Microsoft Research is to design efficient learning algorithms to speed up supervised query expansion. My work at Bing is query parsing, which converts users' question queries into structured formats so that we can search knowledge graph to retrieve answers.