Inspiration The inspiration for this project came from the common "Bureaucratic Maze" that students face. Often, a student with a simple problem is bounced from one department to another because they don't know who is responsible for what. We wanted to build a "Single Window" system where the student doesn't need to understand the university's hierarchy—the system understands the student.
What it does SmartResolve AI is an intelligent grievance redressal platform that acts as a bridge between students and university administration.
AI Routing: Students submit complaints in plain English or Hindi. The system uses the Gemini 1.5 Flash API to automatically categorize the issue (e.g., Hostel, Finance, Academic) and assign a priority level.
Smart Dashboard: Department heads receive pre-sorted, summarized tickets, allowing them to focus on high-priority issues first.
Escalation Matrix: If a student is not satisfied with a resolution, the matter is automatically escalated to higher authorities (Dean/Registrar).
PVC Oversight: A dedicated executive dashboard for the Pro-Vice-Chancellor to monitor university-wide resolution rates and department performance.
How we built it The project is built on a robust and scalable tech stack:
Backend: Python/Django was used for the core business logic and database management.
AI Intelligence: We integrated the Google Gemini 1.5 Flash API to perform real-time text analysis, sentiment detection, and automated summarization.
Database: PostgreSQL handles the relational data for students, staff, and grievance tickets.
Frontend: A responsive, mobile-first UI built with Bootstrap 5 to ensure students can report issues from their phones.
Logic: We implemented a custom Escalation Algorithm that triggers when a user provides negative feedback on a resolved ticket.
Challenges we ran into Reliable Data Formatting: Getting the AI to consistently return structured JSON data that the backend could parse was tricky. We solved this through rigorous Prompt Engineering.
Language Variance: Students often use "Hinglish" (a mix of Hindi and English). Ensuring the AI accurately categorized these mixed-language inputs required specific instruction tuning.
Escalation Logic: Designing a flow that moves a ticket between different authority levels without losing the history of the complaint was a complex database design challenge.
Accomplishments that we're proud of Zero-Knowledge Entry: We successfully created a system where a student needs zero knowledge of university departments to get their problem to the right person.
Automated Prioritization: Our AI can distinguish between a "broken fan" (Medium Priority) and a "safety emergency" (High Priority) instantly.
Accountability: We built a transparent system where every action taken by the staff is logged and visible to the higher management.
What we learned Prompt Engineering: We learned how to use LLMs (Large Language Models) as "Logic Routers" rather than just chat interfaces.
Full-Stack Integration: Gained deep experience in connecting a modern AI API with a traditional web framework like Django.
User-Centric Design: Realized that in administrative tools, the goal is to reduce the "clicks" a user has to make to reach a solution.
What's next for Single Window Student Grievance System Voice-to-Ticket: Integrating speech-to-text so students can record their grievances via voice notes.
Predictive Analytics: Using historical data to predict recurring issues (e.g., predicting water shortages in specific hostels before they happen).
WhatsApp Integration: Building a WhatsApp bot interface so students can check their ticket status via a simple message
Built With
- gemini
- mern
Log in or sign up for Devpost to join the conversation.