Hey Dona! AI Course Assistant

Inspiration

Course registration is often a stressful process for students, involving long queues, clashes, and complicated forms. We envisioned an intelligent agent—"Hey Dona!"—that could handle all of this through a simple voice conversation. Inspired by the possibilities of LLMs and conversational AI, we created a solution that could revolutionize academic interactions.

What it does

Hey Dona! is a smart, voice-activated AI assistant that:

Helps students register for courses.

Checks for scheduling conflicts and prerequisites.

Provides real-time feedback and confirmation.

Integrates with academic databases (mocked for demo) to simulate realistic actions.

How we built it

We used:

Speech recognition APIs to interpret user commands.

A large language model (LLM) to understand intent.

A rule-based logic system for course validation and scheduling.

A lightweight Python backend and simple UI to simulate course registration.

Challenges we ran into

Balancing free-form voice input with strict academic rules.

Creating mock academic data that accurately represented real systems.

Maintaining accuracy in LLM interpretation across varied student queries.

What we learned

How to blend NLP with rule-based systems for structured workflows.

The potential of AI agents in streamlining administrative tasks.

Voice UI design principles and natural language error handling.

What’s next

Integrate with real university databases.

Add multilingual support for wider accessibility.

Expand to broader tasks like assignment reminders, attendance, and academic advising.

Built With

• api)
• backend
• course
• database)
• demo
• face
• flask
• for
• google
• hugging
• llm)
• mocked
• python
• speech-to-text
• sqlite
• transformers
• youtube