I worked with Jesse Nunez, Madhav Viswesvaran, Ishaan Avinash, Calais Waring, and Anna Lin to develop SlangSwitch, a slang-to-formal translator that also predicts regional variation, age group, sentiment, and the overall percentage of slang in a user's input. The project combined neural style transfer with demographic-focused linguistic analysis to explore how informal language varies across generations and regions. We used Styleformer, built on top of Hugging Face transformers, to translate slang and informality into more formal text. Beyond translation, the tool incorporated regional identification through web-scraped slang dictionaries, age prediction using frequency-based slang lists for >18, 18–29, and 30+, sentiment analysis through TextBlob and NLTK, and percentage analysis using the OpenAI API to quantify slang density.

Within the team, I designed the front-end interface, including the visual layout of SlangSwitch, and I created the poster and presentation materials for the Data Science UCSB Project Showcase. I also handled the web-scraping of regional slang and built the regex-based matching system that linked user input to region-specific terminology. Our project won 2nd place at the Data Science UCSB Project Showcase.