Xinyu Tang Email & Phone Number

Centreville, Virginia, United States

Los Angeles, California, United States

Developed and implemented a Python-based web application for cryptocurrency data analysis and trading, utilizing Flask for the backend and Streamlit for the frontend, which provided AI-driven insights and trend visualizationEngineered with team to build a multi-layered system architecture with a Data Collection layer (sourcing data from business news, social media, and whitepapers), a ETL layer, and an Applications layer, leveraging a Retrieval-Augmented Generation (RAG) approach to enhance AI-driven insightsEstablished and managed an AWS S3 bucket for future database storage, utilizing AWS CLI and IAM for secure setup. Integrated this with the system’s Data Processing Layer to facilitate seamless data handling and storageExecuted a comprehensive data cleaning script in the ETL pipeline by using Python, including offensive language filtering, language detection, word count filtering, duplicate content and ads removal. Processed data was then uploaded to an AWS S3 bucket, ensuring a clean, ready-to-analyze dataset. Convert whole ETL pipeline into automation process in Python

Monterey Park, California, United States

Extract data entries from a Telco company's website backend in the Data Analysis Department, focusing on customer behavior analysis. Utilized Python to investigate the relationship between user activity and stickiness, leading to actionable insights that supported the development of strategies to improve customer retention and overall satisfactionImplemented Logistic Regression, Random Forest, and Support Vector Machine (SVM) algorithms using Python with scikit-learn, and conducted model tuning and evaluation using ROC curves. Achieved an 80.41% accuracy with Random Forest, which was suggested as the most accurate model for future predictions. Automated the prediction process in Python, with code running in the terminal, providing the most accurate model each timeUsed matplotlib and seaborn to visualize data and understand feature relationships related to user churn. Analyzed algorithmic outcomes using pandas and numpy, identifying "OnlineSecurity", "TechSupport", and "MonthlyCharges" as the critical factors influencing churn. Provide insights with the operations and customer support teams

Cambridge, Massachusetts, United States

Collected and compiled road safety data for 11 Southeast Asian countries from multiple international sources (WHO, UN ESCAP, CIA, IRF, and World Bank), and processed institutional framework data and HDI values for comprehensive analysisDeveloped Python code using NumPy and Pandas for Min-Max, Vector, and Max normalization methods; implemented the Entropy method for indicator weighting, performing data normalization, and producing computational results over multiple yearsApplied Entropy-embedded TOPSIS and RSR methods in Python using Scikit-learn for road safety performance evaluation and country ranking; conducted robustness analysis using PCA (Scikit-learn); grouped countries based on performance metricsDe-constructed and visualized road safety indicators using radar charts with Matplotlib and Seaborn, presenting insights and trends over four years; de-composed normalized indicator values for grouped countries to facilitate policymaker decision-makingCollaborated with a team and mentors to produce the published article, "Measuring road safety achievement based on EWM-GRA-SVD: a decision support system for APEC countries"

Master Of Science - Ms, Business Analytics And Risk Management

I chose to specialize in Business Analytics and Risk Management for my graduate studies with the intention of delving deeper.

Bachelor Of Science - Bs, Chemistry

Throughout my university career, laboratory work has been a significant component of my experience. In the process of acquiring knowledge.