** Reliability Data System – Data Engineer • Process 1B records of data per day from data centers to provide data views for reliability engineers. • Lead 2 interns to complete data pipelines to visualize data for reliability engineers. ** Quality Management System – Data Engineer • Increased correctness rate of server components by 120% by leading data collection projects to get aligned with the data in internal databases. • Reduced runtime of data pipelines by 80% via replacing Hive with Spark. In the same time, the cost is reduced by 60% with transiting from Hadoop cluster to serverless Spark cluster. • Lead 9 Indian contractor to complete service migration to meet Microsoft compliance.

** Home Network Security – Data Engineer • Reduced 90% time of reports from 1B security events every day. This helps marketing and sales people in Japan, Singapore and Australlia to find opportunities to improve business. • Visualized the relationship between security events for thread experts with word2vec and t-SNE. ** Network Behavior Analysis Project – Data Scientist • Developed a machine learning model to recognize IoT devices based on 30 billion records of netflows via Spark in Scala and Python. • Reached a 90% accuracy rate in identifying periodic network behaviors of IoT devices with a statistical model.

** Yield Improvement Project – Data Engineer and Data Scientist • Processed the big volume of data (6TB per day) to maintain a data warehouse for machine learning projects. • Reduced the out-of-control rate by 30% via a statistical model. • Reduced scrapping rate by 80% with homemade anomaly detection algorithms. • Reduced 80% time to find key factors of yield rates via data visualization and statistics ** Big Data Solutions – Data Engineer • Digest 6TB data per day by building an on-premise big data solution via Scala, Spark and Hive. • Reduced 95% implementation time of machine learning algorithms via R, MPI, Hive and Spark. ** Weekly Productivity Improvement Program – Leader • Developed R packages to reduce reinventing the wheels and increase productivity. • Taught writing clean and performant codes to data scientists and data engineers. • Organized study groups to share knowledge of machine learning and statistics with colleagues.

** Main role • Decreased data processing time by 80% via R and MongoDB to process millions of records of data per day. • Got a 40% lowered RMSE in imputing missing values with home-made machine learning than other methods.