Smart Assistance and Complaint Portal

Campus Info & Issues — Solved Smarter

Comment

Inspiration

We noticed two major problems on campus: students struggle to find simple information (like exam dates or fees) buried in circulars, and facility complaints (like broken fans) often get ignored or lost in informal chats. We wanted to build a single, intelligent platform to solve both issues instantly.

What it does

It is a unified web portal that serves as a 24/7 campus companion. For Students: It provides an AI Chatbot ("Ask AI") that gives instant, accurate answers from official college documents. It also allows them to report facility issues and track their resolution status in real-time. For Admins: It provides a dashboard to view, manage, and resolve student complaints efficiently.

How we built it

1.Frontend: Built with HTML, CSS, and Vanilla JavaScript as a Single Page Application (SPA) for fast, smooth performance without page reloads.

2.Backend: Developed using Python Flask to handle API requests, authentication, and logic.

3.Database: Used PostgreSQL (via Supabase) to securely store user credentials and complaint tickets.

4.AI Engine: Integrated Google Gemini API (1.5 Flash) using RAG (Retrieval-Augmented Generation). The AI reads from a text-based "Knowledge Base" and the live database to generate context-aware answers.

Challenges we ran into

AI Hallucinations: Initially, the AI gave generic answers. We fixed this by implementing RAG, forcing it to read only from our provided college_data.txt file.

API Rate Limits: We faced 429 Quota Exceeded errors with experimental models. We solved this by switching to the stable gemini-1.5-flash model.

Database Latency: We dealt with frontend timeouts when the database was slow. We resolved this by adding robust error handling and loading states in the UI.

Accomplishments that we're proud of

1.Smart Context: Successfully making the AI "aware" of live database issues (e.g., it knows if a projector is broken).

2.Full Stack Integration: seamlessly connecting a custom frontend, a Flask backend, a cloud database, and an AI service into one working product.

3.User Experience: Creating a clean, dark-themed UI that is intuitive for both students and admins.

What we learned

1.RAG Architecture: We learned how to "teach" an LLM specific knowledge without retraining it.

2.Full-Stack Debugging: We mastered tracking errors from the browser console down to the server logs.

3.Prompt Engineering: We discovered that giving the AI a "persona" and strict rules is crucial for getting consistent, professional responses

What's next for Smart Assistance and Complaint Portal

1.Multi-Format Support: Upgrading the AI to read directly from PDFs and Images (circulars).

2.Notifications: Adding Email or SMS alerts when an admin resolves an issue.

3.Voice Support: Enabling students to ask queries via voice commands for better accessibility.

Built With

Share this project:

Updates

Submission history