School of Technology and Computing
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
Event details
Date: Thursday, June 11th, 2026
Time: 4:30PM-5:30PM PT
Contact: ravidivakarreddy@cityuniversity.edu
Advisors
DR. Sam Chung
Dean,
School of Technology & Computing
School of Technology & Computing
Michael Lahman
Professor,
School of Technology & Computing
(CS497 & DS687)
School of Technology & Computing
(CS497 & DS687)
TaeJin Kim
Professor,
School of Technology & Computing
(CS687 & AI687)
School of Technology & Computing
(CS687 & AI687)
Sivakumar Visweswaran
Professor,
School of Technology & Computing
(CS687)
School of Technology & Computing
(CS687)
Session Chairs
David Vo
Divakar Reddy Ravi
Harrison Le
Clark Ngo
Akshita Modi
Giridhar Gopi Krishna
Yashasree Duggiempudi
Thao Bui
Vishal Krishna Kumar
Mai Hung
Namrata Vijay Mali
Sumit K C
Sunidhi Hegde
Presenters
Ayush Kumar Gupta
AI-Powered Care Coordination Agent: A Semantic Framework for Chronic Care Management (CCM)
Course: CS 687
Click here to access the Video Recording
Description: The research focuses on transforming passive EHR platforms into proactive, state-aware clinical assistants using Agentic AI, the Model Context Protocol (MCP-AI), Fast Healthcare Interoperability Resources (FHIR), and semantic web technologies. The proposed system uses a multi-agent architecture to monitor patient timelines, detect care gaps such as missed follow-ups or unmanaged hypertension, and generate context-aware summaries for clinicians. By integrating longitudinal reasoning, semantic interoperability, and human-on-the-loop governance, the framework aims to reduce clinician administrative burden, improve chronic disease management, and support outcome-based healthcare models like the 2026 CMS ACCESS initiative.
Sai Mani Ritish Upadhyayula
Cerebellar-Inspired Spiking Neural Network Controllers for Jerk-Limited Robot Motion.
Course: CS 687
Click here to access the Video Recording
Description: A research project proposing a cerebellar-inspired spiking neural network using LIF neurons and STDP to learn jerk-limited robot motion profiles, compared against a quintic polynomial baseline on CartPole using metrics such as peak jerk, endpoint error, computational cost, and post-stop oscillation.
Akarsh Lakshmana
AIOps Self-Healing Kubernetes Platform
Course: TBA
Click here to access the Video Recording
Description: AI-driven self-healing infrastructure on AWS EKS. Detects anomalies with an ensemble ML model, predicts failures before they happen, automatically remediates issues, and surfaces everything in Grafana.
Harrison Le
LitConnect: Expanding Learning Management System with Social Connectivity and AI Tutoring
Course: CS 687
Click here to access the Video Recording
Description: Learning Management Systems (LMSs) are widely used in higher education to manage course materials, assignments, assessments, grades, and academic communication. However, traditional LMS platforms often remain content-centered and administrative, offering limited support for social interaction, peer networking, and personalized learning assistance. To address this limitation, this research presents LitConnect, a local proof-of-concept educational technology platform that integrates LMS functionality, social learning features, and a local AI tutor called PAIT.
Sara Verkiyani
AI-Powered Agentic Travel Planning and Recommendation System
Course: CS 687
Click here to access the Video Recording
Description: An AI-powered travel planning system that utilizes agentic AI techniques to generate personalized travel itineraries, recommend accommodations and activities, optimize routes, and support travel-related decision-making. The system integrates user preferences, external travel data, and large language models to provide intelligent and adaptive travel planning assistance.
Name: Sukhmani Thukral
An Agentic AI System for Community Epilepsy Awareness
Course: CS 687
Click here to access the Video Recording
Description: A communication system designed for epilepsy awareness, using agentic AI to provide accurate, accessible, and timely responses to community questions.
Name: Im Rothpanhaseth
An AI-Powered Multi-Agent Collaboration Platform for Student Project Management
Course: CS 687
Click here to access the Video Recording
Description: CollabAgent is an AI-powered web and mobile collaboration platform designed to support student capstone project management. The system uses multiple AI agents to assist with task management, team coordination, knowledge sharing, advisor feedback, and project tracking. Features such as intelligent task management, Retrieval-Augmented Generation (RAG), and real-time collaboration tools help improve communication, productivity, and workflow efficiency. The platform brings students and faculty advisors together in a single place to manage capstone projects more effectively throughout the project lifecycle.
Name: Namrata Vijay Mali
A Zero-Trust Semantic Middleware for Model Context Protocol (MCP) Compliance
Course: CS 687
Click here to access the Video Recording
Description:Today’s AI assistants can connect to real tools balances, transfers, tickets not just chat, and once you’re logged in the system often trusts whatever the assistant tries to do until you sign out. This capstone asks: does the next action still match what the user wants right now? Attackers can hide instructions in emails, pages, or files the assistant reads and steer it toward unintended moves say, wiring money during what should be a harmless lookup. Sentinel is a prototype guardrail on that tool path: it stores the session goal in plain language and, before risky tools run, uses meaning-based comparison (not only keywords) between the planned action and that goal blocking misaligned calls and logging why then evaluates against a safe simulated banking benchmark using Control, a simple text-rule baseline, and Sentinel to show trade-offs between stopping harmful actions and mistaken blocks.
Name: Wu Peini
Predictive Inventory and Procurement Optimization for Cross-Border Furniture Retail
Course: CS 687
Click here to access the Video Recording
Description: A predictive inventory and procurement optimization application for a small cross-border furniture retailer.
Name: Zhorobekova Zeinep
AI-Powered Resume Analyzer and Job Match Scoring System
Course: CS 687
Click here to access the Video Recording
Description: This project is an AI-powered web application that analyzes resumes and compares them with job descriptions using GPT-4o. The system provides a semantic match score, identifies matched and missing skills, and generates personalized recommendations to improve resume-job alignment.
Name: de Oliveira Panta Cordeiro Jheyne
AI-Powered Product Feedback Insights
Course: CS 687
Click here to access the Video Recording
Description: This project leverages AI to transform product feedback into actionable insights using rating-based sentiment analysis and data exploration. Through an interactive Streamlit dashboard, it helps product managers identify trends, prioritize improvements, and make data-driven decisions.
Name: Fade Mitch
Creation of a Spectral Monitoring System for Satellite RF Data
Course: CS 687
Click here to access the Video Recording
Description:My capstone project created a Convolutional Neural Network (CNN) classifier to classify RF signals from satellite downlink data. The classifier preprocesses the data applying data modifications such as doppler corrections, amplitude scaling, gaussian noise to the dataset. The CNN classifier uses a 1 Dimensional ResNet architecture to identify modulation types of satellite signals.
Name: Gilbert Jack
Title: TBA
Course: CS 687
Description: TBA
Name: Gudivada Likhitha Lakshmi
Security Risks and Guardrails in Large Language Models: A Practical Analysis of LLM Vulnerabilities and Safety Mechanisms
Course: CS 687
Click here to access the Video Recording
Description:This project focuses on improving the security, safety, and reliability of Large Language Models (LLMs) by implementing guardrail-based protection mechanisms. The project examines major LLM vulnerabilities such as prompt injection attacks, hallucinations, unsafe content generation, and sensitive data leakage, which can impact the trustworthiness of AI systems in real-world applications. A secure LLM workflow prototype was developed using AWS cloud services including Amazon API Gateway, AWS Lambda, Amazon Bedrock, Bedrock Guardrails, AWS CloudWatch, and AWS CDK. The system performs input validation, content filtering, risk classification, output moderation, and logging/monitoring to detect and reduce unsafe interactions before responses are returned to users. The overall goal is to demonstrate practical security mechanisms that can support safer deployment of AI-powered applications in cloud environments.
Hiltzman David Sebastian
QuoteSentry: ML-Powered Compliance for Secure Engineering Quotes
Course: CS 687
Click here to access the Video Recording
Description:QuoteSentry is a self-hosted, ML-powered procurement platform that integrates export classification screening, drawing change detection, and supplier quote anomaly detection into a single compliance-aware sourcing workflow for ITAR and EAR-regulated organizations.
Krisnawan Andrew
AI Generated Image Detection Model
Course: CS 687
Click here to access the Video Recording
Description: Developing a machine learning-based digital forensic framework to distinguish between authentic photography and synthetic media generated by modern AI models like GANs and Diffusion models.
Mingsorn Siraphat
Development of an AI-Based Mock Interview Mobile Application with Automated Feedback
Course: CS 687
Click here to access the Video Recording
Description:My capstone project focuses on developing a mobile-first AI-based mock interview application that helps students and job seekers practice role-specific interview questions, submit text-based responses, receive automated feedback, and review session history. The goal is to make interview preparation more accessible, practical, and effective.
Nelson Jeff Kenneth
Dialect Bridge — Viennese Dialect to Standard High German Translation System
Course: CS 687
Click here to access the Video Recording
Description: Dialect Bridge is a Natural Language Processing (NLP) project that demonstrates and compares two approaches for translating between Viennese Austrian Dialect and Standard High German. The project combines a rule-based lexical substitution technique with a Large Language Models (LLMs) using few-shot prompting to improve translation quality for this low-resource dialect. A human-curated “Gold Set” dataset is used for system development and translation quality evaluation. Translation performance is measured using the industry-recognized metrics BLEU, chrF, and COMET. The system aims to preserve semantic meaning and regional linguistic characteristics while serving as a proof-of-concept for future translation systems designed for low-resource languages and dialects.
Sridhar Sai Shruthi
A Decision-Centric Agentic System for Autonomous Customer Churn Intervention in Software-as-a-Service (SaaS) Businesses
Course: CS 687
Click here to access the Video Recording
Description: ChurnAlert AI is an AI-powered system that helps SaaS businesses reduce customer churn by turning risk predictions into actionable interventions. It automatically identifies at-risk customers, explains why they might leave, and recommends personalized retention strategies for Customer Success Managers to review and approve. The system uses an agentic workflow where AI agents autonomously monitor customer behavior, reason over risk signals, and generate intervention recommendations. Built as a full-stack application, it keeps humans in control of every decision through a review and approval step before any action is taken.
Vanani Vidhiben Ashokbhai
Smart Roommate Matching System
Course: CS 687
Click here to access the Video Recording
Description: The Smart Roommate Matching System is a full-stack web application designed to help students and professionals find compatible roommates based on lifestyle preferences, budget, daily routines, and personal habits. The system uses cosine similarity techniques to calculate compatibility scores and generate intelligent roommate recommendations. The application is developed using Next.js for the frontend, FastAPI for the backend, PostgreSQL for database management, and JWT authentication for secure user access. The project aims to improve shared living experiences by reducing roommate conflicts and providing a more personalized and efficient roommate matching process.
Chambers Brayden Thomas Q
Frame Time Anomaly Detection with LSTM Autoencoding
Course: CS 497
Click here to access the Video Recording
Description: A machine learning recognition software for identify abnormalities in frame-rate consistency from video game graphical performance data.
Collins Arlena
Secure Cloud-Based File Storage System
Course: CS 497
Click here to access the Video Recording
Description: This project focuses on improving cloud security in AWS environments by using a layered security approach. It examines how tools like CloudTrail, S3, and IAM can be used together to monitor activity, control access, and protect stored data. The project highlights common risks, such as misconfigurations and unauthorized access, and demonstrates how proper security practices can reduce these threats. Overall, the goal is to show practical ways to strengthen cloud security and improve data protection in real-world systems.
Fisher Douglas
Using Machine Learning to Predict Social Media Engagement
Course: CS 497
Click here to access the Video Recording
Description: Social media engagement can be challenging to forecast, and many content creators depend on an iterative approach when sharing content. This project investigates how machine learning techniques can be utilized to analyze social media data and accurately predict engagement metrics, including likes, comments, and shares.
Huynh Minh
A Retrieval-Augmented Generation Framework for Enhancing the Muddiest Point Pedagogical Technique
Course: CS 497
Click here to access the Video Recording
Description: This project introduces an AI-driven tutoring system that integrates Retrieval-Augmented Generation (RAG) with Socratic scaffolding to transform traditional formative classroom assessments into an active, real-time “thinking coach.” Rather than offering direct answers to students’ points of confusion, the system uses course-grounded materials to prompt guided inquiry, cultivating independent critical thinking and metacognition.
Jackson Dillon
Mapping Host Forensics to MITRE Attack for Incident Response
Course: CS 497
Click here to access the Video Recording
Description: Project focuses on building a host based forensic investigation framework that maps APT29 ATT&CK techniques to Windows forensic artifacts within a controlled lab environment.
Jackson Riley
Defending the Modern Network: Integrating Zero Trust Architecture, AI-Driven Intrusion Detection, and Encrypted Traffic Analysis for Next-Generation SOC Operations
Course: CS 497
Click here to access the Video Recording
Description: This capstone research addresses a critical gap in enterprise cybersecurity by proposing a unified defense framework for Security Operations Centers that integrates three proven but separately deployed technologies. Zero Trust Architecture enforces identity-based access control across perimeter-less environments. AI-Powered Intrusion Detection uses ensemble machine learning to detect behavioral anomalies with explainable outputs. Encrypted Traffic Analysis surfaces threats hidden inside TLS sessions without requiring payload decryption. No existing peer-reviewed study has combined all three into a single operational SOC model. This research fills that gap through a systematic literature review and a prototype analyst interface called CipherWatch, which demonstrates how the three layers work together to give SOC analysts clearer, more contextual, and more actionable threat intelligence.
La Brake Sean
A Proof of Concept for Integrated Audio-Visual Inventory and Event Management
Course: CS 497
Click here to access the Video Recording
Description: A Proof of Concept for Integrated Audio-Visual Inventory and Event Management explores how centralized synchronization and real-time operational visibility can improve coordination during live AV event operations. The project addresses common gaps between inventory tracking systems and event planning workflows by demonstrating a simplified synchronization prototype capable of supporting mobile operational updates and centralized dashboard visibility. Using lightweight cloud-based tools as a proof of concept, the project validates how synchronized operational workflows may improve equipment visibility, issue-state tracking, and coordination across multiple simultaneous event environments while supporting the feasibility of more advanced integrated AV management systems in future implementations.
Latif Emma
Smart Locker System
Course: CS 497
Click here to access the Video Recording
Description: This project is investigating how the locker system generates codes and delivers code retrieval to customers. The focus is on understanding the algorithms used for code generation and their security implications. Additionally, the research aims to evaluate the efficiency of the retrieval process and identify any potential areas for improvement.
Ramirez Jr Joel
Leveraging AI for Small Business Automation
Course: CS 497
Click here to access the Video Recording
Description: My Capstone project focuses on how AI assisted tools and cloud-based business platforms can help improve operational efficiency for small businesses. Using a real world case study involving SynPros Inc., a small plumbing business based in Sacramento, California, the project evaluates workflow inefficiencies related to invoicing, scheduling, customer communication, and payment tracking. The project explores practical and affordable solutions using technologies such as QuickBooks Online, Pagecloud, Google Workspace, and AI assisted tools to streamline operations, improve customer engagement, and reduce administrative workload.
San Vijay
Bridging the Gap: AI-Driven Virtual Assistance for Mitigating Analyst Fatigue in Security Operations
Course: CS 497
Click here to access the Video Recording
Description: Modern Security Operations Centers face an alert paradox in which massive volumes of telemetry cause severe analyst burnout and alert fatigue. This capstone presents an artificial intelligence assistant acting as a force multiplier to bridge this gap. Utilizing an advanced generation pipeline with Wazuh, n8n, and Ollama, the system transforms unstructured logs into comprehensive attack narratives. It maps threats using the MITRE framework and generates actionable Sigma rules. Uniquely, its Socratic interface explains the reasoning behind alerts to upskill junior analysts. Empirical testing revealed that the prototype reduced manual triage time by over 30% while maintaining a 90% expert agreement rate. By processing critical threats faster than human workflows can, this automated solution enhances operational efficiency and mitigates professional analyst burnout.
Thoreson Danielle Dakota
SipSync: A Smart Hydration Tracking System
Course: CS 497
Click here to access the Video Recording
Description: SipSync, a smart hydration tracking system designed to help users monitor their daily water intake in a simple and interactive way. The goal of this project is to encourage healthier hydration habits by combining hardware and software into one easy to use system. SipSync uses a Raspberry Pi 5, load cell sensors, and a custom interface to measure and display water consumption data in real time. The system is designed to provide accurate tracking, personalized hydration recommendations, and a user friendly experience that can help individuals stay more aware of their daily health habits. This project also explores how affordable smart technology can be used to create practical health focused solutions for everyday life.
Wright Tyler C
Augmented, Not Automated: Why AI Won’t Replace the Cybersecurity Professional Unless You Let It
Course: CS 497
Click here to access the Video Recording
Description: This capstone examines the growing risk of passive AI adoption in cybersecurity, where professionals increasingly accept automated threat detection outputs without meaningful human review. Drawing on research in machine learning, human factors, and explainable AI, the project argues that skill erosion and automation complacency pose a serious organizational risk. The deliverable is a Human-in-the-Loop Security Dashboard that trains a classifier on real network intrusion data, uses SHAP to explain every model decision to the analyst, and routes low-confidence detections to a human review queue rather than acting on them automatically. The goal is to demonstrate empirically that human oversight produces measurably better security outcomes than full automation.
Dang Quoc Vuong
Advancing XAI RelEx(2.0): Integrating Llama 3.1-Instruct and E5 for Advanced Research
Course: DS 687
Click here to access the Video Recording
Description: This capstone project presents RelEx 2.0, an XAI-enhanced Information Retrieval framework designed to address the transparency gap and risk of hallucinations in modern black-box neural search engines. The system integrates high-dimensional dense vector retrieval using E5-Base-v2 and coordinates deep semantic reasoning through the Llama 3.1 Large Language Model. RelEx 2.0 extracts contextually accurate keyphrases, dynamically shifts search queries using an Extended Rocchio Algorithm, and supports human-in-the-loop relevance feedback via an interactive UI, enabling highly assessable, user-adaptive search workflows that are mathematically anchored by LIME and SHAP frameworks.
Hagin Michael Terrell
Credit Card detection using Machine Learning
Course: DS 687
Description: This project explores the use of machine learning for credit card fraud detection. Various classification models and imbalance-handling techniques are used to identify fraudulent transactions while improving accuracy and reducing false positives.
Lee Hannah
Developing a Security and Privacy Checklist for Smart Home Security Cameras
Course: CY 665
Click here to access the Video Recording
Description: This capstone develops and evaluates a practical security framework for consumer smart home security cameras by identifying common vulnerabilities, mapping them to real-world attack scenarios, and translating those findings into an evidence-based, plain-language checklist that non-technical homeowners can use to harden their devices throughout the lifecycle (before purchase, during setup, and ongoing use).
Amarnath Vidhyalakshmi
A Decision-Aware Multi-Modal Financial Intelligence Framework with Directional Edge Forecasting, Predictive Throttling, and Regime-Adaptive Execution Control.
Course: AI 687
Click here to access the Video Recording
Description: Financial forecasting systems frequently struggle to maintain robustness under volatile market regimes, noisy information flows, and unstable execution environments, limiting their practical applicability in real-world financial decision systems. This paper presents a decision-aware multi-modal financial intelligence framework that integrates Temporal Fusion Transformers (TFT), directional edge forecasting, event-driven sentiment modeling, predictive throttling, and regime-adaptive execution control within a unified live forecasting and paper-trading architecture. The proposed framework combines heterogeneous financial information sources, including historical market dynamics, cross-asset relationships, macroeconomic indicators, volatility and stress signals, point-in-time fundamentals, event-driven features, and FinBERT-based financial sentiment representations. Rather than optimizing solely for raw return prediction, the framework employs a directional edge-aware learning objective designed to estimate forward trading opportunity while supporting downstream execution decisions and risk-sensitive allocation policies. To enhance operational robustness, the system introduces predictive throttling and hierarchical execution controls that dynamically regulate exposure under elevated uncertainty, volatility shocks, and adverse market conditions. A sequential decision layer further incorporates confidence-aware trade filtering, allocation constraints, horizon-adaptive execution logic, and live paper-trading validation prior to final trade approval. Extensive walk-forward experiments conducted across major U.S. equities under transaction-cost-aware evaluation demonstrate stable directional forecasting performance, improved risk-adjusted behavior under predictive throttling, and resilient execution characteristics across changing market environments. The proposed framework establishes a practical foundation for adaptive AI-driven financial intelligence systems capable of jointly integrating forecasting, uncertainty estimation, execution control, sequential evaluation, and real-time deployment within a unified research and production-oriented architecture. The live evaluation further incorporates realized learning memory to analyze trade-level outcomes, utility-score calibration, directional edge realization, and production-gate effectiveness over time.
Boren Michael Reese
Web-Augmented LLM Pipelines for Verified Food Recognition and Macronutrient Estimation: A Multi-Phase Accuracy Evaluation
Course: AI 687
Click here to access the Video Recording
Description: Manual food logging fails because users abandon it within weeks, and current vision-language models still hit error rates too high for clinical or athletic use. This capstone presents IronFeed, a tiered nutrition resolution system that routes each meal to the strongest available evidence source (barcode lookup, USDA FoodData Central, chain restaurant menus, or web search) and labels every output with its provenance tier so users can calibrate trust. A three-pipeline benchmark on 30 meals shows IronFeed reaching 87.8% accuracy versus 78.4% for an LLM with web search and 70.3% for a plain LLM, with a follow-up constrained resolver pilot lifting accuracy on home-cooked plates from 41% to 76%.
Schedule
Main Room & Session 01
Session 02
Session 03
Session 04
Session 05
Session 06
Session 07
Session 08
Session 09
Session 10
Session 11
Session 12
Session 13
Main Room & Session 01
| 4:30 PM – 4:35 PM | Welcoming session _____________________________________________ | Main Advisor: Dr. Sam Chung Main Chair: Divakar Reddy Ravi MSTeams link: Session 1 |
| 4:35 PM – 4:55 PM | Presenter 1: Ayush Kumar Gupta Topic: AI-Powered Care Coordination Agent: A Semantic Framework for Chronic Care Management (CCM) _____________________________________________ | Chair: Divakar Reddy Ravi Advisor: Sivakumar Visweswaran Presentation: 15 Minutes Q&A: 5 minutes |
| 4:55 PM – 5:15 PM | Presenter 2: Sai Mani Ritish Upadhyayula Topic: Cerebellar-Inspired Spiking Neural Network Controllers for Jerk-Limited Robot Motion. ____________________________________________ | |
| 5:15 PM – 5:35 PM | Presenter 3: Akarsh Lakshmana Topic: AIOps Self-Healing Kubernetes Platform _____________________________________________ | |
Session 02
| 4:35 PM – 4:55 PM | Presenter 1: Harrison Le Topic: LitConnect: Expanding Learning Management System with Social Connectivity and AI Tutoring _____________________________________________ | Chair: Harrison Le Advisor: Sivakumar Visweswaran Presentation: 15 Minutes Q&A: 5 minutes MSTeams link: Session 2 |
| 4:55 PM – 5:15 PM | Presenter 2: Sara Verkiyani Topic: AI-Powered Agentic Travel Planning and Recommendation System _____________________________________________ | |
| 5:15 PM – 5:35 PM | Presenter 3: Sukhmani Thukral Topic: An Agentic AI System for Community Epilepsy Awareness _____________________________________________ |
Session 03
| 4:35 PM – 4:55 PM | Presenter 1: Im Rothpanhaseth Topic: An AI-Powered Multi-Agent Collaboration Platform for Student Project Management _____________________________________________ | Chair: Namrata Vijay Mali Advisor: Sivakumar Visweswaran Presentation: 15 Minutes Q&A: 5 minutes MSTeams link: Session 3 |
| 4:55 PM – 5:15 PM | Presenter 2: Namrata Vijay Mali Topic: A Zero-Trust Semantic Middleware for Model Context Protocol (MCP) Compliance _____________________________________________ | |
| 5:15 PM – 5:35 PM | Presenter 3: Wu Peini Topic: Predictive Inventory and Procurement Optimization for Cross-Border Furniture Retail _____________________________________________ |
Session 04
| 4:35 PM – 4:55 PM | Presenter 1: Zhorobekova Zeinep Topic: AI-Powered Resume Analyzer and Job Match Scoring System _____________________________________________ | Chair: Clark Ngo Advisor: Sivakumar Visweswaran / Taejin Kim Presentation: 15 Minutes Q&A: 5 minutes MSTeams link: Session 4 |
| 4:55 PM – 5:15 PM | Presenter 2: de Oliveira Panta Cordeiro Jheyne Topic: AI-Powered Product Feedback Insights _____________________________________________ | |
| 5:15 PM – 5:35 PM | Presenter 3: Fade Mitch Topic: Creation of a Spectral Monitoring System for Satellite RF Data _____________________________________________ |
Session 05
| 4:35 PM – 4:55 PM | Presenter 1: Gilbert Jack Topic: TBA _____________________________________________ | Chair: Vishal Krishna Kumar Advisor: Taejin Kim Presentation: 15 Minutes Q&A: 5 minutes MSTeams link: Session 5 |
| 4:55 PM – 5:15 PM | Presenter 2: Gudivada Likhitha Lakshmi Topic: Security Risks and Guardrails in Large Language Models: A Practical Analysis of LLM Vulnerabilities and Safety Mechanisms _____________________________________________ | |
| 5:15 PM – 5:35 PM | Presenter 3: Hiltzman David Sebastian Topic: QuoteSentry: ML-Powered Compliance for Secure Engineering Quotes _____________________________________________ |
Session 06
| 4:35 PM – 4:55 PM | Presenter 1: Krisnawan Andrew Topic: AI Generated Image Detection Model _____________________________________________ | Chair: Thao Bui Advisor: Taejin Kim Presentation: 15 Minutes Q&A: 5 minutes MSTeams link: Session 6 |
| 4:55 PM – 5:15 PM | Presenter 2: Mingsorn Siraphat Topic: Development of an AI-Based Mock Interview Mobile Application with Automated Feedback _____________________________________________ | |
| 5:15 PM – 5:35 PM | Presenter 3: Nelson Jeff Kenneth Topic: Dialect Bridge — Viennese Dialect to Standard High German Translation System _____________________________________________ |
Session 07
| 4:35 PM – 4:55 PM | Presenter 1: Smith Aaron J. Topic: TBA _____________________________________________ | Chair: Sumit KC Advisor: Taejin Kim Presentation: 15 Minutes Q&A: 5 minutes MSTeams link: Session 7 |
| 4:55 PM – 5:15 PM | Presenter 2: Sridhar Sai Shruthi Topic: A Decision-Centric Agentic System for Autonomous Customer Churn Intervention in Software-as-a-Service (SaaS) Businesses _____________________________________________ | |
| 5:15 PM – 5:35 PM | Presenter 3: Vanani Vidhiben Ashokbhai Topic: Smart Roommate Matching System _____________________________________________ |
Session 08
| 4:35 PM – 4:55 PM | Presenter 1: Thoreson Danielle Dakota Topic: SipSync: A Smart Hydration Tracking System _____________________________________________ | Chair: David Vo Advisor: Taejin Kim Presentation: 15 Minutes Q&A: 5 minutes MSTeams link: Session 8 |
| 4:55 PM – 5:15 PM | Presenter 2: Chambers Brayden Thomas Q Topic: Frame Time Anomaly Detection with LSTM Autoencoding _____________________________________________ | |
| 5:15 PM – 5:35 PM | Presenter 3: La Brake Sean Topic: A Proof of Concept for Integrated Audio-Visual Inventory and Event Management _____________________________________________ |
Session 09
| 4:35 PM – 4:55 PM | Presenter 1: Jackson Dillon Topic: Mapping Host Forensics to MITRE Attack for Incident Response _____________________________________________ | Chair: Akshita Modi Advisor: Michael Lahman Presentation: 15 Minutes Q&A: 5 minutes MSTeams link: Session 9 |
| 4:55 PM – 5:15 PM | Presenter 2: San Vijay Topic: Bridging the Gap: AI-Driven Virtual Assistance for Mitigating Analyst Fatigue in Security Operations _____________________________________________ | |
| 5:15 PM – 5:35 PM | Presenter 3: Jackson Riley Topic: Defending the Modern Network: Integrating Zero Trust Architecture, AI-Driven Intrusion Detection, and Encrypted Traffic Analysis for Next-Generation SOC Operations _____________________________________________ |
Session 10
| 4:35 PM – 4:55 PM | Presenter 1: Fisher Douglas Topic: Using Machine Learning to Predict Social Media Engagement _____________________________________________ | Chair: Giridhar Gopi Krishna Advisor: Michael Lahman Presentation: 15 Minutes Q&A: 5 minutes MSTeams link: Session 10 |
| 4:55 PM – 5:15 PM | Presenter 2: Huynh Minh Topic: A Retrieval-Augmented Generation Framework for Enhancing the Muddiest Point Pedagogical Technique _____________________________________________ | |
| 5:15 PM – 5:35 PM | Presenter 3: Collins Arlena Topic: Secure Cloud-Based File Storage System _____________________________________________ |
Session 11
| 4:35 PM – 4:55 PM | Presenter 1: Latif Emma Topic: Smart Locker System _____________________________________________ | Chair: Mai Hung Advisor: Michael Lahman Presentation: 15 Minutes Q&A: 5 minutes MSTeams link: Session 11 |
| 4:55 PM – 5:15 PM | Presenter 2: Ramirez Jr,Joel Topic: Leveraging AI for Small Business Automation _____________________________________________ | |
| 5:15 PM – 5:35 PM | Presenter 3: Wright Tyler C Topic: Augmented, Not Automated: Why AI Won’t Replace the Cybersecurity Professional Unless You Let It _____________________________________________ |
Session 12
| 4:35 PM – 4:55 PM | Presenter 1: Dang Quoc Vuong Topic: Advancing XAI RelEx(2.0): Integrating Llama 3.1-Instruct and E5 for Advanced Research _____________________________________________ | Chair: Sunidhi Hegde Advisor: Sivakumar Visweswaran and Michael Lahman Presentation: 15 Minutes Q&A: 5 minutes MSTeams link: Session 12 |
| 4:55 PM – 5:15 PM | Presenter 2: Hagin Michael Terrell Topic: Credit Card detection using Machine Learning _____________________________________________ | |
| 5:15 PM – 5:35 PM | Presenter 3: Lee Hannah Topic: Developing a Security and Privacy Checklist for Smart Home Security Cameras _____________________________________________ |
Session 13
| 4:35 PM – 4:55 PM | Presenter 1: Amarnath Vidhyalakshmi Topic: A Decision-Aware Multi-Modal Financial Intelligence Framework with Directional Edge Forecasting, Predictive Throttling, and Regime-Adaptive Execution Control. _____________________________________________ | Chair: Yashasree Duggiempudi Advisor: Taejin Kim Presentation: 15 Minutes Q&A: 5 minutes MSTeams link: Session 13 |
| 4:55 PM – 5:15 PM | Presenter 2: Boren Michael Reese Topic: Web-Augmented LLM Pipelines for Verified Food Recognition and Macronutrient Estimation: A Multi-Phase Accuracy Evaluation _____________________________________________ | |
| 5:15 PM – 5:35 PM | Presenter 3: NA Topic: NA _____________________________________________ |