SCHOOL OF TECHNOLOGY AND COMPUTING

School of Technology and Computing
Applied Research Symposium Fall 2025

Hosted by CityU’s Center for Cybersecurity Innovation and School of Technology and Computing (STC), the Applied Research Symposium is open to faculty, professionals, and students from any discipline, university, or professional organization.  At the end of each quarter, any students who would like to share their projects including capstone courses will present their outcomes. This symposium will provide opportunities for students, researchers, and practitioners to discuss the influence and impact of the applied computing on the future of our planet and our society

Join Symposium

Event details

Date: Thursday, Dec 11th, 2025
Time: 4:30PM-5:30PM PT
Contact: lenghiemminhhieu@cityuniversity.edu

Advisors

DR. Sam Chung
Dean,
School of Technology & Computing
(CS497) (CS680)
Michael Lahman
Michael Lahman
Professor,
School of Technology & Computing
brian maeng
DR. Brian Maeng
Professor,
School of Technology & Computing
MORGAN ZANTUA
Professor,
School of Technology & Computing
TaeJin Kim
Professor,
School of Technology & Computing
(CS687)
Sivakumar Visweswaran
Professor,
School of Technology & Computing
(CS687)
DR. Patrick Offor
Professor,
School of Technology & Computing
(CY665)
Jemell Garris
Professor,
School of Technology & Computing
(CS497)
Session Chairs
Harrison Le
Ashwin Shastry Paturi
ClarkNgo
Clark Ngo
Rajdeep Singh
Megha Narendra Simha
Sumit K C
Namrata Vijay Mali
Presenters
Ana Karina Balcazar Sanabria

Ana Karina Ba;cazar Sanabria

Solution Architecture Multi-Agent Customer Service Assistant

Course: CS687

Description: This project develops a multi-agent customer service assistant in Microsoft Copilot Studio to simulate a telecom helpdesk. Using agentic AI, specialized agents handle issues like weak signals, and billing errors, while a central assistant combines their inputs into a single response. The system shows how agentic AI can automate routine support tasks, improve efficiency, and deliver faster, more accurate help. It also offers practical experience in building multi-agent conversations, integrating knowledge bases, and modeling real business processes. Overall, the project connects technical innovation with business value, emphasizing responsible AI, system integration, and clear communication in enterprise settings.

Khoa Nguyen

Quiz Me.AI: An AI-enabled Quiz Mobile App

Course: CS497

Description: How can you develop a smart mobile learning app that uses artificial intelligence to help with explaining your answer?

Chao-Kai Lin

MCP-Based RAG Multi-Agent Resume Screening System

Course: CS687

Description: Establish an intelligent resume screening system based on RAG (Retrieval-Augmented Generation) and MCP (Model Context Protocol) architecture, which can automatically analyze the match between resumes and job descriptions, and provide structured analysis reports and candidate rankings.

Rajdeep Singh Golan

Punjabi NLP Evaluation: Sentiment Analysis and Summarization with Multilingual LLMs

Course: CS687

Description: This capstone project evaluates the performance of multilingual large language models (mBERT, mT5, and LLaMA) on Punjabi text. It includes dataset collection from OSCAR and Kaggle, preprocessing for Gurmukhi and Romanized scripts, model fine-tuning using Hugging Face, and evaluation using Accuracy, F1-score, and ROUGE metrics to assess linguistic understanding and text generation quality.

Tyler Kepler

Automated Open Source Provenance Analyzer 

Course: CS 687

Description: Leveraging Machine Learning to Assess Geographic and Organizational Risk in Open-Source Software Dependencies

Honorine Ndom Ndzah

Multi-Agent System in Smart Manufacturing

Course: CS687

Description: An LLM-enhanced multi-agent system that integrates Large Language Models (LLMs), Long Short-Term Memory (LSTM)-based predictive analytics, and the Agent Development Kit (ADK) to enable richer, scalable, and context-aware decision-making in a smart factory environment.

Zook Gango

Securing IoT Medical Devices Through a Zero Trust Framework

Course: CY665

Description: This project investigates systemic vulnerabilities across medical device lifecycles–including authentication weaknesses, insecure communication pathways, and insufficient update mechanisms–to evaluate their impact on patient safety and data integrity.  By synthesizing current research, regulatory guidance, and industry best practices, the project develops a structured approach for implementing Zero Trust principles in healthcare environments. The overarching goal is to provide a scalable, evidence-based model that strengthens organizational resilience while enabling safer, more reliable IoMT ecosystems.

Keerthi Gajjela

Smart Crop Recommendation System Using Machine Learning

Course: CS687

Description: My capstone project develops an AI-driven smart farming system that analyzes soil nutrients, weather conditions, and leaf images to recommend the best crop, suitable fertilizer, and detect plant diseases. Using machine learning and CNN models, the system supports farmers with accurate, data-driven agricultural decisions.

Kina Anh Bui

Anh Bui

AI-Based Lead Scoring and Generative Marketing Recommendation System 

Course: CS687

Description: My capstone develops an AI-based Lead Scoring and Generative Marketing Recommendation System that combines machine-learning models with Generative AI to improve how businesses qualify leads and create personalized marketing content. The project includes data preprocessing, predictive modeling, evaluation, and a Flask demo that automates both lead scoring and marketing message generation. This solution shows how AI can enhance marketing efficiency and support data-driven decision-making.

Brandon Foster

Cyber Forensics in a Quantum Computing World

Course: CY665

Description: A look at the progress that could be made in Cyber forensic investigations with the help of Quantum Computing

Chandrahasa Munagala

E-Commerce Logistics Optimization Agent 

Course: CS687

Description: This project develops an intelligent logistics optimization system for e-commerce operations. It integrates machine learning, demand forecasting, and OR-Tools–based routing to assign orders efficiently, reduce delivery time, and minimize operational costs. The system predicts demand, selects the best warehouse, and generates optimal delivery routes and uses Groq’s LLM inference to answer delivery-related queries.

Emmanuel Kensese

The Impact of Artificial Intelligence on Toxic Waste Management and Disposal: Advancing Efficiency, Safety, and Environmental Sustainability

Course: CS 497

Description: This capstone project presents a simplified AI-driven hazardous waste management prototype designed for warehouse and industrial environments. The system uses a convolutional neural network (CNN) to classify waste items as hazardous or non-hazardous using image-based input, with real-time confidence scores and probability visualization to support human-in-the-loop decision-making. A simulated IoT monitoring module demonstrates environmental awareness by using temperature, VOC, and bin-fill data. The project emphasizes safety, transparency, and modular design to support future expansion to full sensor integration and predictive analytics.

Neha Arora

Trip-Mate: Multi-Agent AI for Smart Travel Planning

Course: CS687

Description: Trip-Mate is an AI-driven multi-agent platform that provides personalized, adaptive, and efficient travel planning. Specialized agents manage itineraries, flights, accommodations, and activities, coordinated through LLM-based reasoning for a seamless, conversational user experience.

Braedon David Merlo

Model Context Protocol (MCP) Security Sanitizer Agent

Course: CY665

Description: The MCP agent enforces a zero-trust architecture, securing agentic AI by applying bi-directional sanitization to neutralize malicious prompt injection attacks delivered via PDFs. Once the JSON has been sanitized, the agent leverages a trust token system for data integrity, performing cryptographic hashing and re-hashing. This process protects against manipulation and data exfiltration while also enabling efficient PDF caching.

Jocelyn E Jeriah

Balancing Popularity and Discovery on Spotify: Addressing Bias in Music Recommendation Algorithms

Course: CS497

Description: A study on how popularity bias shapes Spotify’s recommendations and a prototype that gives users control over the balance between similarity and discovery to improve fairness and diversity.

Zandra Nguyen

AI-Powered Mental Health Chatbot

Course: CS687

Description: An AI mental health chatbot that understands emotions and offers supportive, safe, and responsive guidance.

Keerthi Turakapalli

AI Agent for Loan Approval Using Multi-Agent Behavioral Analysis and Real-Time Fraud Detection

Course: CS687

Description: This project develops an intelligent fraud-detection system that analyzes customer behavior and identifies suspicious loan applications in real time. Using a multi-agent architecture, the system detects anomalies, sends automated alerts, and blocks high-risk applications before approval. The solution enhances accuracy, reduces fraud risk, and supports faster, more secure loan-processing decisions.

Ruojie Hao

AI-Powered Engineering Internship at AriesView

Course: CS680

Description: This capstone presents my internship at AriesView, a Boston-based AI real estate technology company. My work focused on full-stack engineering, OCR and LLM-based document processing, and the development of retrieval-augmented generation (RAG) pipelines for commercial real estate workflows. Through this internship, I strengthened my engineering skills, improved UI/UX consistency across enterprise modules, and learned how AI tools are integrated into real-world SaaS products following agile development practices.

Sai Sathyak Nallamalli

Medisecure AI - Smart Insurance Assistant

Course: CS687

Description: This project is an AI-driven insurance management system that allows users to search patient information, interact with a chatbot to query insurance details, and generate personalized PDF reports. It integrates Django for backend logic, MongoDB for patient storage, ReportLab for document generation, and a custom rule-based AI agent for answering natural-language questions. The system makes healthcare record access simple, conversational, and automated — enhancing user experience with intelligent insights and document export features. 

Amber Giza

Predicting Future Homelessness and Chronic Homelessness Rates with RNNs

Course: DS687

Description: With the usage of HUD data and other government data on homeless people and chronically homeless people, can future rates for both groups be predicted with RNNs? If so, between LSTMs and Elman RNNs, which framework is more accurate at predicting those future rates?

Vidhyalakshmi Amarnath

Simulation Benchmarks for Decentralized Agents in Water Distribution

Independent Research (Submitted to Springer)

Description: This research investigates the use of Multi-Agent Reinforcement Learning (MARL) to enhance the resilience of water distribution networks. The study focuses on how autonomous agents can optimize resource allocation during critical scarcity events, providing a scalable framework for smart infrastructure management.

Schedule

Contact Us

lenghiemminhhieu@cityuniversity.edu