Developed a Human-AI collaborative system using Django and SQLite to assist instructors in creating high-quality study questions, enhancing instructors' question creation efficiency by 25%.Designed user interfaces that allow instructors to manipulate students' previous submissions in custom-built editors to create new questions.Leveraged GPT-4 model to simplify, formalize, and contextualize texts, as well as generate questions aligned with instructors' specific criteria, enhancing the efficiency of content creation.Implemented an auto-save feature to ensure continuous information preservation throughout the content creation process.Optimized dynamic rendering speed by 40\% through lazy loading and caching mechanisms.

Researched on the impact of YouTube’s removal of the dislike button on users’ commenting behaviors, focusing on natural language processing (NLP) techniques to interpret user sentiment.Applied Python’s NLTK and TextBlob libraries to perform tokenization and sentiment analysis on over 400 scraped user comments, achieving an accuracy rate close to 90%.Improved sentiment analysis accuracy by 7% through text data cleaning and identifying noise and outliers in the datasets.