About

George Liu is a highly skilled and experienced data scientist with over 8 years of relevant experience. He has a unique skill set that combines machine learning, statistics, engineering, and product management, making him an expert in designing and building AI solutions that solve business problems. He is currently the Head of Data Science at Chatter Research, where he leads an AI team of 6 members and manages AI development end to end, including data labeling, data quality, data tooling product management, ML modeling, deployment, and ML performance monitoring and improvement. George has a proven track record of delivering significant efficiency and capability improvements for both startups and corporate organizations. He specializes in data science solution building, machine learning (ML), deep learning (DL), natural language processing (NLP), A/B testing, product management, project management, communication, presentation, and team building. George is also an experienced mentor, having conducted weekly teaching sessions on various data science topics for Great Learning and Udacity, where he helped students acquire data science skills using his unique capability to turn complicated concepts into easy-to-understand vivid examples. George has a diverse educational background, having completed a Master of Business Administration (MBA) from the University of Toronto - Rotman School of Management and several online courses and specializations in NLP, deep learning, statistics, data science, and machine learning from various universities and platforms such as deeplearning.ai, Coursera, and Udacity. He holds a Bachelor of Engineering (B.Eng.) in Automotive Engineering from Shenyang Aerospace University. George has a deep understanding of various technologies and tools such as Keras, Hyperparameter, CNN, QA, Search, Salesforce, Postgres, Mobile, and others. He has worked on several notable projects, including building and productionizing a multiclass, multilabel text classification deep learning model that achieved a 50% increase in topic prediction accuracy, while reducing model training time by over 90%. He also designed, built, and deployed Topic Performance Forecaster, an RNN-based deep learning model that predicts future topic performances and the resulting NPS score, allowing retailers to identify top topics to work on to boost NPS.