AskMAV

Posted on by Steven McDermott

Team Name

AskMav Development

Timeline

Summer 2025 – Fall 2025

Students

  • Robert Darragh – Computer Engineering
  • Suhas Gillipelli – Computer Science
  • Mohammed Shoiab – Computer Science

Abstract

AskMav is a Generative AI based assistant that is intended to help University of Texas at Arlington students with questions and concerns that would otherwise be difficult to seek out or must be discussed with a particular individual, for example an academic advisor. The product is in the form of a web application so that it is portable and easy to access. For more permanent installations, a kiosk will be utilized. This kiosk will allow students to swipe their ID so that the system can easily recognize them.

Background

Universities often rely on websites, help desks, or office staff that may not be available 24/7, can be difficult to navigate, or require students to already know where to look. Important details about advising, enrollment, financial aid, scholarships, or campus resources are often spread across many websites or offices, making it difficult to know who to ask or where to look. Askmav addresses this problem by centralizing information in an easily accessible kiosk or webapp.

Project Requirements

  • User Session Management
  • Natural Language Understanding
  • Kiosk Enclosure
  • Web Application Hosting
  • Administrative Dashboard and Logging

Design Constraints

  • Training on a dataset with university-specific terms and common student queries.
  • Kiosk enclosure is minimally obstructive and easily accessible while adhering to accessibility guidelines.
  • Compliance with university data security policies and handle peak traffic during high-demand periods.
  • Tools and environment to maintain the physical kiosk must be kept within the university’s IT infrastructure and any remote access must use encrypted channels.
  • The kiosk must comply with data privacy and ethical AI usage guidelines.

Engineering Standards

  • ADA Accessibility Requirements: Physical accessibility for public-use kiosks.
  • WCAG 2.1 AA: Digital accessibility for touchscreen interfaces and web content.
  • OWASP ASVS: Security guidelines for protecting web applications and APIs.
  • ISO/IEC 27001 – Information Security Management Systems (ISMS): Requirements for managing information security risks, including access control, data handling, logging, and incident response.
  • ISO 9241-110: Design principles for user interface, including clarity, feedback, error tolerance, and user control.

System Overview

The AskMav system has a three-layer architecture that consists of the kiosk layer, frontend/client layer, and backend/LLM layer. The kiosk layer includes the Raspberry Pi hardware and peripherals such as the touchscreen, microphone, PIR sensor, and card reader; it provides the physical interface and local device control. The frontend/client layer delivers the user interface, allowing students to ask questions and view responses through an accessible web-based experience. The backend/LLM layer processes user requests and piping them into our chosen API, then sending the generated response back to the frontend. These three layers work together to deliver our interactive information kiosk.

Results

AskMav application successfully carried out the mission of being an assistant to answer university-specific questions in a pre-packaged, deployable kiosk. The system allows student chat sessions to persist between logins so they can easily refer to previous conversations.

Future Work

Future of this project could involve training an AI specifically on UTA advisement resources instead of using only publicly available data. Another area that can be improved is the aesthetic of the kiosk design and how the peripherals are mounted.

Project Files

Project Charter
System Requirements Specification
Architectural Design Specification
Detailed Design Specification
Poster

References

Steven McDermott

More Posts