Portfolio

Industry Projects at SHEIN

Fashion Attributes Extraction and Enhance Application with CLIP

Leverage Large Multimodal Models like GPT4, Gemini to recognize and categorize fashion attributes from competitors’ products, alongside employing CLIP for fashion product recall based on top search keywords. Achieving average 90%+ accuracy in 500+ attributes lables.

Enhanced Product Categorization with RAG

This project aimed to revolutionize the way product titles are matched with category trees by integrating NLP techniques with Retrieval-Augmented Generation (RAG). Achieving 93%+ accuracy in multilables categorization.

Enhanced Fashion Design with Stable Diffusion

This project utilizes Stable Diffusion to innovate fashion design and presentation. Key features include:Trendy Redesigns, Style and Fabric Recoloring and Background Replacement.

Competitor Intelligence Framework with Big Data Governance

As Big data engineer, development big data governance framework with Hadoop for competitors’ product data from web crawlling, around tens of billions daily. Covering from AWS original s3 data files into CV embedding vector system application.

Event Tracking Management and User Analysis OLAP Platform

Implement event tracking management system to manage events life cycle. Build user behavior analytics modules with OLAP techs and advanced data structure like linked list and directed graph to visualize data. Improve query efficiency by 30%, monthly average users 230.

Education

University of Wisconsin - Madison

Aug.2024 - Jun.2026 (Expected)
Master of Science in Computer Science
- Advanced Natural Language Processing, Machine Learning

South China University of Technology

May.2019 – Apr.2021
Bachelor of Accounting
Courses:
- Statistic Pinciple, Database Management, Data Modeling, Probability Theory, Calculus, Linear Algebra

Coursera Certificates

Page template forked from evanca