Codex Community Hackathon Seoul (3rd Place)

Our team placed third at the Codex Community Hackathon Seoul with EFFICA. The event brought students together at AWS Korea for a day of building in teams that combined implementation and problem-definition roles. Our political science teammate's concern about political bias and declining political efficacy among young people gave us a reason to build before we chose the technology.

EFFICA connects reporting about the same issue so readers can compare shared facts and differences in framing. LLM analysis examines political bias and sensationalism alongside reader evaluations. A perspective map connects the overall distribution to an individual article's analysis, while links to the original reporting let readers inspect the evidence themselves.

I led the full-stack implementation and helped translate the analysis criteria into the product. Article analysis runs asynchronously through the OpenAI Responses API, with reasoning and confidence stored alongside reader assessments. The interface keeps those assessments distinct: a bias score does not determine factual accuracy, and analysis confidence is not source credibility. The aim is to help people explore unfamiliar perspectives and form their own judgment.