• University/College: University of California, Berkeley
• Major: Business Analytics
• Minors: NA
• Start Date: 20 August 2024

• University/College: University of California, San Diego
• Major: Data Science
• Minors: NA
• Start Date: 20 September 2023

• University/College: Tongji University
• Major: General Engineering
• Minors: NA
• Start Date: 04 September 2018

Data Science Intern, User Growth Team of Ele.me E-commerce Platform

Developed and integrated user labels from multiple data sources by utilizing SQL to process 900 million records and optimized the efficiency through resource allocation and query tuning, improving processing speed by 30% Addressed user label distortion issues by innovating new features and employing an XGBoost model to achieve 89.3% accuracy and 86% MAC recall, conducted offline evaluation using key metrics, and built a data dashboard Innovated user consumption segment using new labeling methods, supported by A/B testing, lifting order frequency by 3%, which directly informing operational strategies launched in September 2024 Collaborated with product, operation, and BI teams in developing user labels, achieving a daily usage of over 1 billion API calls and enabling seamless integration with more than 600 downstream tables Analyzed the growth potential of users by utilizing data visualization and conducting detailed metric drill-downs

Financial Data Science Intern, Financial Engineering Group

Developed ETL pipelines with Python and SQL to process 1 billion financial records from multiple data providers and implemented multiple partitioned tables in DWS and ADS layers to extract features and support downstream analytics Utilized PyTorch to construct CNN models on stock candlestick charts to predict stock trends, achieving a precision of 61% and improving the effectiveness of an established trading strategy by 5% Innovated a CNN model incorporating six time-series feature operators for automated feature engineering to predict stock returns, resulting in 22.3% annualized return, which was presented at the CFRI & CIRF Joint Conference 2023 Developed a sentiment classification model by fine-tuning the FinBERT model on manually labeled corporate earning calls, achieving an average F1 score of 0.88 in a three-class setup

• LGBTQ: no
• Ethnicity: Asian
• Disability: no
• Veteran: no
• First-generation College Student: yes
• Pell Grant Eligibility: no
• Work Authorization: yes
• Visa Sponsorship: no