Ranking and Recommendation ML

• Designed, built and launched an internal tool to automatically query training data, perform feature selection, leverage sampling & training strategies, train ML model with threshold tuning, and generate rules based on human gave initial conditions; reduced human involved percentage by 80%, increased rule generation speed from 1 week to 1 day; succeeded in 93.33% precision.• Developed and integrated semi-supervised learning and active learning extensions for this tool, providing various choices on different use-cases.• Built and Deployed custom kernel with this tool integrated, supporting fast easy-to-use libraries and APIs.

• Developed an Intelligent Code Review Tool using Code Token Analysis, Word2Vec, RNN, Attention Mechanism to automatically locate buggy code snippets in PHP files; succeeded in 91.8% bug location accuracy.• Deployed Intelligent Code Review Tool on the online code Submission System of Tencent ULink Project; increased peer code review efficiency by 3 times and reduced human resources by 50%.• Implemented an online chat web system with both Python & PHP using WebSocket and Redis based on Django Web Framework, providing real-time communication for different web applications.• Organized and hosted the Half-year Company Party of Tencent IEG Value Added Service Department in Shanghai.