CampusBuddy

Inspiration

Campus communities often rely on forums and messaging groups that lack moderation, structure, and reliability, especially during high-stress periods like exams or placement seasons. We wanted to build a safe, engaging, and intelligent student platform where AI not only helps students but is also observable, reliable, and secure. CampusBuddy was inspired by the idea that AI in education must be trusted, and trust comes from visibility. Campusbuddy is a fully serverless platform.

What it does

CampusBuddy is an AI-powered student engagement platform where students can:

1. Post academic doubts and get structured discussions
2. Share notes and resources
3. Discover campus events and opportunities
4. Interact in a toxic-free, moderated environment
5. Interactive games. Powered by Google Gemini 2.5 Flash, the platform:
6. Automatically summarizes long posts
7. Categorizes content (Notes, Events, Queries, Opportunities)
8. Detects and blocks toxic or unsafe content in real time What makes CampusBuddy unique is that every AI decision is observable using Datadog.

How we built it

1. Frontend built with React 19 and Tailwind CSS, featuring a festive glassmorphism UI.
2. Firebase Authentication for secure Google Sign-In.
3. Supabase for real-time post and user data.
4. Firebase Cloud Functions to handle AI workflows.
5. Google Gemini 2.5 Flash API from Google AI Studio for moderation, summarization, and categorization.
6. Datadog RUM + Custom Actions to monitor AI latency, failures, and security signals. We instrumented Gemini calls with Datadog custom events to create a Glass-Box AI system.

Challenges we ran into

1. Monitoring AI behavior, not just frontend clicks.
2. Handling Gemini rate limits and failure scenarios.
3. Designing real-time toxicity detection without reviewing user content manually.
4. Balancing AI automation with a smooth, zero-friction user experience.

Accomplishments that we're proud of

1. Built a fully observable AI lifecycle (latency, failures, security).
2. Implemented a Real-Time Toxicity Radar.
3. Created alert rules like: If AI failure rate > 5% in 5 minutes → trigger alert.
4. Delivered an enterprise-grade observability approach in a student platform.
5. Designed a premium, festive, responsive UI.

What we learned

1. AI systems need monitoring just like microservices.
2. Observability increases trust in AI-driven platforms.
3. Datadog Custom Actions can bridge frontend UX and backend AI logic.
4. Educational platforms benefit greatly from silent, background AI enhancements.

What's next for CampusBuddy

1. Admin dashboards for AI risk trends
2. Campus-level analytics for student engagement
3. Multilingual AI moderation
4. Expansion to inter-college communities
5. Advanced anomaly detection on AI outputs

Built With

• antigravity
• datadog-custom-actions
• datadog-rum
• firebase-authentication
• firebase-cloud-functions
• firebase-firestore
• firebase-hosting
• git
• github
• google-ai-studio
• google-gemini-2.5-flash-api
• html
• lucide-react
• npm
• react-19
• react-router-dom
• supabase
• tailwind-css
• web-vitals