Competition Overview
- Placement: 1st of 87 teams in Memetics track.
- Prize: Best Memetics Hack, sponsored by Calcifer Computing.
Project Links
Technologies Used
- Python
- FastAPI
- Next.js
- TypeScript
- Gemini
- Grok
Details
I traveled out to JHU this year mostly looking to continue my hackathon research and meet some cool people! While I definitely accomplished that (even getting to chat with Cory Levy, founder of Z Fellows!), I also managed to pick up 1st place in one of the event's main prize tracks.
For this project I worked alongside Hao and Mahmudul, whom I met via Discord the day before the event. We had a really fun time building CapClipper, which parses huge volumes of X posts to analyze subtrends, with the goal of identifying coordinated botting and misinformation campaigns. Working with a corpus of 395 million posts, we ran sentiment analysis and LLM-assisted labeling to identify trends within GLP-1 weight-loss discussions. This yielded a ton of insights, like a heavily botted campaign of grey-market peptide distributors.
In addition to this project, I got kinda bored at 2am and made a SECOND project where I applied CapClipper's methods to 26,000 comments from my old Minecraft YouTube channel. You can check out that project here.