1. "Understanding User Behavior in a Cross-domain Scenario" - Establish a query suggestion system analyzing users’ behavior of when they changed browsing domains - Design and implement a critical ingestion pipeline to process over 5 TB of e-commerce data from Verizon Media using Spark and Pandas - Develop novel embedding mechanism which collects time-/cross- domain features and conducted experiments via deep learning framework (RNN-based, Attention) using TensorFlow and PyTorch 2. "Query Forecasting: Identifying Critical Learning Queries" - Analyzed data from Verizon Media’s e-commerce platform, which has millions of active users, and constructed a pipeline to analyze the textual embedding using Python, Shell scripts, and established modeling tools - Proposed and constructed a quasi-attention deep learning framework with TensorFlow, aiming to perceive customers’ purchasing intention. Simulation results verified the superiority of the proposed method 3. "Foresight Research and Its Application in Financial Technology Industry" - Utilized Twitter content, technical analysis, and Natural Language Processing (NLP) to analyze US stock market - Implemented machine learning algorithms to classify market sentiment using Python and SQL