\section*{Project Overview}
\subsection*{Inspiration} The project was inspired by the problem of unreliable, unverified, and unmoderated course reviews at IIT Delhi. Existing platforms lack authenticity, leading to misinformation, bias, and bandwagon effects.
\subsection*{What It Does} We built a platform that enables: \begin{itemize} \item Kerberos-verified anonymous reviews \item Two-layer AI-based moderation system \item Trust scoring for users and reviews \item Anti-bandwagon voting with hidden vote counts \end{itemize}
\subsection*{How We Built It} The system was developed using: \begin{itemize} \item \textbf{Frontend:} Next.js 16, React 19 \item \textbf{Language:} TypeScript \item \textbf{Styling:} TailwindCSS \item \textbf{AI Integration:} Claude API \end{itemize}
\subsection*{Challenges} \begin{itemize} \item Tuning moderation boundaries to balance censorship and openness \item Designing robust anti-manipulation mechanisms \item Implementing a fail-open architecture without compromising safety \end{itemize}
\subsection*{Accomplishments} \begin{itemize} \item Developed a novel two-layer moderation pipeline \item Implemented hidden vote counts to reduce bias \item Designed structured review formats for higher signal quality \end{itemize}
\subsection*{What We Learned} \begin{itemize} \item Prompt engineering as a core design tool \item Principles of structured anonymity \item Trade-offs involved in fail-open system design \end{itemize}
\subsection*{What’s Next} \begin{itemize} \item Persistent database integration \item Real Kerberos SSO implementation \item Live AI-generated summaries \item Professor response channel \item Expansion to multiple campuses \end{itemize}
