Research Interests
Cybersecurity
Network security, penetration testing, and secure systems design
Machine Learning
AI-powered secure systems and predictive analytics
CS Education
Innovative approaches to teaching computer science
Smart Grid Security
Privacy and security for advanced metering infrastructure
Recent Projects
ProgressLM: General Reward Model for Embodied Agents
Can AI look at a single photo of a task in progress and estimate how far along it is? We tested 14 vision-language models and found that most struggle with this kind of temporal reasoning—they can describe what they see, but not gauge how much work remains. Inspired by how humans intuitively judge progress, we fine-tuned a lightweight 3B-parameter multimodal LLM that learns human-like progress reasoning and matches or beats much larger models. This work pushes embodied intelligence closer to understanding “where I am” in a task—a key capability for autonomous long-horizon decision-making.
Students: Qian Chengxuan
SMTest Audits: Comprehensive Cybersecurity Assessments
In this project, we developed and conducted a multi-phase cybersecurity engagement, consisting of a formal Cybersecurity Audit and a targeted Penetration Test for local businesses. Following the NIST Cybersecurity Framework, we evaluated administrative and technical controls to identify critical vulnerabilities and compliance gaps within local organizations. The technical assessment involved deep reconnaissance, web application testing, and wireless network exploitation, alongside social engineering simulations via phishing. The resulting summary combines our findings into a clear, prioritized report designed to strengthen the organization's security posture.
Students: Sabrina Barrile, Thomas Tubinas, Michael Judge
nerd_speak
Roll for initiative! nerd_speak is a new esoteric programming language designed to look like a Dungeons and Dragons adventure. This project incorporates ideas of dice rolling and spellcasting to create a new, fun way to program. Created using Java, nerd_speak is an interpreted language meant to add some randomness and creativity into programming.
Students: Theo Osborn
Global News Aggregator – Scoop
A news aggregator that allows users to read global news with AI-assisted research, analysis, and fact-checking. Scoop aggregates updates from 200+ news sources using Real Simple Syndication (RSS) feeds, which are standardized XML text files released by newspapers and similar outlets. Scoop is a full-stack application with Next.JS as the frontend. With the Backend being made with FastAPI, PostgreSQL, and ChromaDB, with a custom-built RSS ingester coded in Rust.
Students: Jordan Green, Tomas Acosta
An LLM-Augmented AutoML Framework for End-to-End Data Analysis
We propose an AutoML-LLM Data Analysis Platform that streamlines end-to-end data analysis workflows by automating data ingestion, cleaning, feature engineering, and model selection. It uses Large Language Models for task detection and research question generation, supporting classification, regression, and clustering. The platform also features automated task type detection, imbalanced data handling, and multi-model benchmarking.
Students: Jie Xu, Xiangchen Jia
Automatic Parking Control System Based on Webots Simulation
This project constructs a virtual experimental environment for automatic parking control using the Webots simulation platform. It integrates manual driving, GPS/compass-based localization, and a trajectory logging module to support data-driven behavioral modeling. The system allows for both real-time manual teleoperation and trajectory-based autonomous replay in a realistic parking lot environment.
Students: Erchen Qu, TianYu Xu, Zy Kang
ASTRO (Arcadia Student Transformed Remote Observatory)
The ASTRO project involves designing and implementing a new control system for the observatory on the roof of Boyer Hall. This system enables users to remotely capture astronomical data from the Observatory and control the decades-old dome.
Students: Paul Broccardi, Cole Baugh, Ian Sharp
Artistic Nexus
Artistic Nexus is a web platform designed to foster a community within the art space by blending traditional art with AI. It allows artists to share their work, network, and utilize AI tools to generate inspiration for future compositions.
Students: Erin MacIver, David Cortez
CatNip
CatNip is a high-performance password-cracking tool designed to expedite the penetration testing process for Red Teams. It utilizes a High-Performance Computing cluster and a front-end GUI to efficiently process and crack password hashes using Hashcat.
Students: Philip Ryskalczyk, Tyler Kuhl, Chloe Fox
Caring for Friends Delivery App
This mobile application, built with Flutter and AWS, helps volunteer drivers track deliveries and find efficient routes in a non-profit organization that helps people in need. It leverages Google Maps API for route optimization and provides a user-friendly interface for managing delivery logistics.
Students: Jason Thomas, Alyssa Quisito, Dylan Alexander
RoundTable Realms
RoundTable Realms is a VR RPG application built in Unreal Engine 5 that combines the convenience of online sessions with the immersion of in-person campaigns. It features customizable 3D environments and an accompanying website for managing character sheets, maps, and campaign materials.
Students: Zachary Albone, Abby Bock, Jamie Conlin, Corey Klevan
DevSecOps in Government Aerospace Defense
This project integrates automated security vulnerability scanning into DevOps workflows by incorporating Black Duck and Coverity into GitLab CI/CD pipelines. It establishes a framework for nightly security scans in Docker environments to ensure real-time vulnerability detection and remediation.
Students: Benjamin Baron
Wy-Fried
This project demonstrates different vulnerabilities of WiFi networks by conducting various attacks to show how devices can be compromised. Using tools like WiFi Pineapple and custom scripts, it highlights the importance of network security and the risks of public WiFi usage.
Students: Dalton Smyrl, Jarrett Williams
LangKnight
LangKnight is an interactive game that challenges users to master English through audio input-based gameplay. It helps players improve their pronunciation, sentence structure, and grammar by using their voice to defeat monsters in various levels.
Students: Morlaye Yansaneh
Developing a Web-Based System for Remote Collection and Analysis of Vehicle Electrical Systems Over CANBus Using Carloop
This project developed a real-time vehicle monitoring system using Carloop (connected to CAN bus in the vehicle), InfluxDB, and Grafana. It collects and visualizes performance data such as speed, RPM, and coolant temperature, allowing for remote analysis and alerts based on predefined thresholds.
Students: Joshua Valle, Alex-Columna Fuentes
Outsourced Worker's Pay Discrepancies in Computer Science Fields
This research investigates pay discrepancies between outsourced and insourced workers in data science and cybersecurity. By analyzing wages relative to minimum wage and cost of living, it highlights significant inequities in global compensation practices.
Students: Xavier Dube
Osmo: an open-source and automated plant care system
Osmo is a low-cost, Raspberry Pi-powered automated plant care system that monitors and adjusts soil moisture levels. It features a web interface for real-time data monitoring and live video streaming, enabling remote plant care from any device.
Students: Thomas Hoy, Brian Torres
CTF Unplugged
Inspired by the Computer Science Unplugged project, a Capture the Flag (CTF) Unplugged was born as an offline cybersecurity competition for high schools. CTF Unplugged introduces participants to cybersecurity competitions by engaging in a fun team-based competition without a need for technical preparation. If you are interested in the project or would like to receive the answers to the CTF Unplugged, please contact me.
Collaboration
I am open for collaboration on various cybersecurity, machine learning, and education projects.
Get in Touch