Jacky Yu, Phd, Mba Email & Phone Number

San Jose, CA, US

Seoul, Seoul, KR

* Optimized ad auction ranking logic via Reinforcement Learning (RL) leading to a 4.4% ad conversion increase. * Introduced a company’s first online TF Query-to-Category model for search retrieval and relevancy control, boosting revenue and GMV by 4% accumulatively through continuous modeling improvement.* Introduced a company’s first online.

San Mateo, California, US

* Leading the effort of refining the recommendation system* Built the ML backend pipeline to power PS5's personalized in-game help system.* Enhanced search engine functionalities, i.e. multi-language suggestion/autocompletion, for PlayStation ecosystem via Machine Learning. * Built knowledge graph, relations between users and contents, of PlayStation.

* Built machine learning models using a gradient boosting algorithm to identify user intention from a given query.* Converted machine learning models into customized POJO files for production.* Reduced 50% model building time by introducing a new machine learning platform that allows training multiple models in parallel.* Automated data labeling processing.

CN

• Developed a real-time pattern recognition algorithm deployed on a smartwatch, which has achieved 99% accuracy in identifying the user intention of reading the watch.
• Prototyped a cloud-based Ad recommender system's REST APIs using MySQL, Jersey, and Tomcat, on AliCloud.
• Developed multi-threading crawlers in Java to obtain all restaurant names and dish names.
• Improved an NLP phonetic matching algorithm which has achieved 98% accuracy in matching user queries with addresses or names.
• Modeled topic classifiers to increase the data filtering efficiency by more than 50%.
• Experience with Spark MLlib’s MapReduce and data pipeline on a Spark VM.

Boston, MA, US

• Applied various supervised learning approaches, SVM, Decision Tree, Bayes, and the ensemble methods to evaluate the features for crime forecasting (ICDM).
• Designed an unsupervised multi-level clustering for feature learning followed by a boosting algorithm to build a rule-based ensemble model that is statistically significantly better than other state-of-the-art models.
• Designed a temporal clustering and association mining approach for learning the spatial-temporal correlations between perceptible water and precipitation clusters from 50 years of climate data.
• Proposed utilizing spatial-temporal correlations to predict severe flooding, which has achieved competitive accuracies and is 10 times faster than the other models (IEEE BigData).
• Experience with using stochastic gradient descent to train the Multimodal Log-Bilinear Language model on a High-Performance Computing Cluster (HPCC) with 10,000+ cores and GPUs.
• Designed an artificial neural network (SOM) to learn the best features from 10,000+ Haar features.

Boston, MA, US

• Computerized the entire service request process to shorten the application lead time for students with disabilities. * Remotely managed and imaged 200+ computers by mastering Ghost Solution Suite.
• Built and maintained My-SQL and Access databases for the Pre-collegiate Program.
• Creating SQL queries and generating monthly reports to track the progress of Pre-collegiate students, and annual reports for the Massachusetts Department of Education.
• Managed Microsoft Active Directory with group policy settings over a working environment with more than 200 domain users.

Newark, California, US

• Led overseas software development teams to develop Linux drivers (using C++) for 3M touch screens.
• Produced project specifications and requirements and adopted the Agile software development cycle.
• Traced the schedule of multiple projects and allocated proper resources using MS Project.

桃園市, TW

• Managed a subsection of the IT department.
• Responsible for training the new IT member in Object-oriented programming (OOP) skills.
• Developed and maintained the HR and payroll system.
• Implemented several internal business systems using Model-View-Controller (MVC) architecture and Oracle database with stored procedures as back-end.
• Adopted object-oriented design and programming to develop a Work-In-Process (WIP) management system for monitoring and controlling manufacturing processes and inventories.
• Implemented Enterprise Resource Planning (ERP) system (Oracle AP) within a global-scale business with over 29,000 employees. Served as a liaison to collaborate between technical and non-technical professionals.

* Developed the communication protocols over RS232 for transferring files to an embedded system. * Designed and implemented data center architecture to remotely collect electricity data in a building. * Developed an algorithm in Assembly to predict the succeeding location of a moving object.

Doctor Of Philosophy (Ph.D.), Computer Science

Mba, Information System

Bachelor Of Engineering (Be), Information And Computer Engineering