AI/ML

Video Understanding Team - Tech Lead to drive the direction of a team of 10~20 engineers. Selected projects: content based video/audio embedding, video/audio categorization, feature engineering for ranking platform, trending videos detection, video graph learning.

Ads Relevance Team - Tech Lead of ads response prediction. Set up roadmap, lead effort of scalable ads ctr prediction offline/online infrastructure overhaul and feature engineering for revenue lift. 5 patents are filed.Content Relevance Team - Design and Implement NLP, machine learning and text mining algorithms and recommendation engine infrastructure. Main technical driver for the Group Discussion related recommendation and People News in Your Network.

As a member at Content Relevance Team of Content and Community Department, design and implement NLP, machine learning and text mining related algorithms. Some key projects include: The first Linkedin in-house Named Entity Extraction and Disambiguation system, which is widely used in products like people news, topic tagging, company news; Junk news classification; Personalization; Content ranking. 1 patent was issued.

Thoora is a start-up internet company on content aggregation, a subsidiary of Rogers Communications. Design and implement innovative machine learning algorithms on computational linguistic and text processing. A set of tools including phrase creation from text stream, feed data cleaning, story-based twitter query creation, named entity extraction and statistical language model, were developed. 4 patents were filed. Mainly develop programs with C++ on Linux.

Create application-driven technology in information retrieval, web mining and machine learning areas as a team lead. NEC is a multi-national company based in Japan. Handle the full cycle of R&D proposal, technology creation, prototype implementation and engineering efforts for products. Developed advanced compact Chinese language model, short message intelligent processing system and entity relationship extraction system. 8 patents were issued or filed. Programming languages include C/C++, Java, Perl, especially C++.

Research on blog data mining as a visiting scholar in database group, the department of computer science and electrical engineering. Advised by Prof. Hector Garcia-Molina.

Developed OCR and handwritten recognition related algorithms and design UI for these products. Hanwang Technology is a leading company in China’s OCR and handwritten recognition market.