AI-Powered Multi-Agent Academic Management System
Overview
CIC Academic Assistant** is an intelligent, AI-powered academic management platform developed for the Canadian International College (CIC).
The system leverages a **Multi-Agent AI Architecture** to provide students and administrators with a natural-language interface for interacting with academic data such as grades, attendance records, academic warnings, and institutional summaries.
Instead of relying on traditional dashboards and static portals, the platform enables users to communicate with the system conversationally through specialized AI agents, each designed to handle a specific academic domain independently and efficiently.
System Architecture
Multi-Agent AI Design
The platform is built using a **Supervisor-Agent Architecture**, where all incoming queries are first analyzed by a central supervisor agent responsible for intent detection and intelligent routing.
Based on the user request, the supervisor dynamically forwards the query to the most relevant specialized AI agent.
Core Agents
| Agent | Responsibility |
| -------------------- | ------------------------------------------------------ |
| `student_supervisor` | Classifies and routes student queries |
| `admin_supervisor` | Handles and routes administrator requests |
| `grades_agent` | Processes grade-related queries and GPA analysis |
| `attendance_agent` | Retrieves and analyzes attendance records |
| `warnings_agent` | Handles academic warnings and risk detection |
| `student_agent` | General student profile and academic information |
| `admin_agent` | Institution-wide summaries and administrative insights |
This modular architecture ensures:
Clean separation of concerns
Scalability and maintainability
Independent agent extensibility
Improved response accuracy
Easier future integration with advanced LLM pipelines
Tech Stack
| Layer | Technology |
| ------------------ | --------------------------------------------------- |
| Frontend / UI | Streamlit with custom CSS theming |
| AI System | Python-based Multi-Agent Architecture |
| LLM Routing | Intent Classification + Rule-Based Supervisor Logic |
| Data Layer | Pandas + CSV-based structured datastore |
| Session Management | Streamlit `session_state` |
| UI Injection | `streamlit.components.v1` + Vanilla JavaScript |
| Styling | Fully custom dark-themed responsive UI |
Authentication & Role-Based Access Control
The system implements a secure dual-role authentication flow:
Student Access
Students can:
Access personal grades
View attendance records
Check academic warnings
Interact with the academic assistant privately
Administrator Access
Administrators have:
Full institutional visibility
Access to all students and departments
Aggregated analytics and summaries
Institution-wide academic monitoring
The platform maintains:
User identity persistence
Role isolation
Session continuity
Full conversational chat history
User Interface & Experience
The project includes a fully customized modern UI built beyond default Streamlit capabilities.
UI Highlights
Custom dark navy + gold design system
Responsive wide-layout dashboard
Sticky navigation bar with live user badge
Dynamic role-aware sidebar
KPI cards for GPA, department, level, and academic status
Custom chat bubbles with agent attribution labels
Persistent sidebar toggle using `MutationObserver`
Smooth conversational AI experience
Data Layer
The `data_loader` abstraction layer centralizes all academic data operations through dedicated loaders:
python
load_student_grades(sid)
load_student_attendance(sid)
load_student_warnings(sid)
load_all_students_summary()
load_dept_summary()
This structure improves:
Data modularity
Maintainability
Query consistency
Future database migration readiness
Key Features
Multi-Agent AI-powered academic assistant
Natural language interaction with institutional data
Intelligent query routing via supervisor agents
Real-time conversational interface
Role-based data isolation
Academic KPI visualization
Persistent chat memory during sessions
Fully customized professional UI
Modular and scalable architecture
Designed For
This platform is designed for educational institutions seeking to modernize student and administrative interaction through conversational AI without replacing existing academic systems.
It acts as an intelligent layer on top of institutional data, transforming traditional academic portals into interactive AI-powered experiences.