Senior Project Summer 2025 Showcase

WifiMon: A Modular WiFi Security Toolkit (Amber Kovacs, Victor Lomba, Hayden Saunders)

WifiMon is a modular WiFi security analysis toolkit designed for educational use on the Raspberry Pi 5. It enables users to observe wireless activity, including access point broadcasts and client probe requests, through a web-based interface. The system supports both live network sniffing (when compatible hardware is available) and offline packet capture analysis using .pcap files. Built with Python, Scapy, and Flask, WifiMon aims to provide a simple and accessible platform for learning network security fundamentals.

Keywords: Wireless Security, Network Security, Packet Sniffing, Raspberry Pi, Scapy

This project, "LocalScene," is a full-stack web application designed to connect musicians, venues, and music enthusiasts by providing a platform for discovering, sharing, and promoting local music events. The system features an interactive map view utilizing Leaflet.js with marker clustering for visualizing concert locations, a comprehensive list view for browsing events, and a secure, authenticated portal for users to submit new show information. The backend, built with Node.js and Express.js, manages concert and user data in a MongoDB database. Key backend functionalities include secure user registration and login with JWT-based authentication and email verification, and robust API endpoints for submitting and retrieving concert data. Concert submissions leverage external geocoding and timezone APIs to ensure accurate location and time information and incorporate duplicate checking mechanisms. The frontend, developed with HTML, CSS, and vanilla JavaScript, dynamically updates content, handles user interactions, and provides a smooth user experience, including sorting concerts by proximity to the user's location. LocalScene offers significant value by centralizing local music event information, fostering community engagement, and supporting the local music scene by making shows more accessible to fans and easier to promote for artists and venues. 

Keywords: Full-stack, Local Music, Web Application, Geolocation, Node.js, MongoDB 

DecepTech is a multimodal deception detection system that integrates vocal and facial analysis to flag instances of deceptive behavior. For the audio channel, it employs Facebook’s wav2vec2 to generate 13-dimensional speech embeddings alongside 20 MFCC coefficients; for the visual channel, it uses dlib’s 68-point facial landmark detector to extract normalized facial vectors. Modality-specific classifiers—including Random Forest, Logistic Regression, and Gradient Boosting—reach up to 89% accuracy on held-out validation subsets (audio: 87% ACC, F1 0.85; video: 83% ACC, F1 0.82). These scores feed into a stacked XGBoost ensemble that fuses both embedding types, achieving 92% accuracy on a separate test set. All models are serialized for lightweight deployment, and evaluation reports include precision, recall, F1-scores, and confusion matrices. While these findings demonstrate the promise of fused multimodal analysis for automated lie detection, additional validation—such as testing on more diverse populations and real-world scenarios—is required before practical deployment. 

Keywords: Deception Detection, multimodal, wave2vec2, facial landmarks 

Despite over a billion potential users, Language Learning games remain an unrealized genre. 40% of the world's population is monolingual, and 50% of the population are games in some capacity, creating a very large market for Language Learning Games. Despite this market however, No game currently incorporates both excellent game design principles and proper language learning methodologies. Resulting in most language learning games just being flashcard-based games, that are at their core unfun, and don't teach more than what users already know. An Ideal game would integrate language learning into core gameplay systems rather than layering them on. That's where our game, Chronicles of the Wordsmith, comes in. We are merging language learning with the fun of a video game.

Keywords: Video Game, Language Learning, Scriptable Objects

PortionPerfect is a website created to help people cook at home using the ingredients they already have. It solves common problems like missing ingredients, wasting food, and struggling to adjust recipes for different serving sizes. The site lets users create an account, add and save their own recipes, and get smart suggestions when they don’t have every item a recipe calls for. It also adjusts the amount of each ingredient based on how much the user has and how many people they want to cook for. PortionPerfect uses Supabase for user login and database management. Recipes, ingredients, and substitutions are stored in a structured database using PostgreSQL. The front end is built with HTML to make the site easy to use on any device. Users only see their own saved recipes, and the system helps them find substitute ingredients when something is missing. This report explains how the site works, including the recipe pages, database design, and how recipe adjustments are made. PortionPerfect encourages smarter, more sustainable cooking by reducing waste and helping people make the most of their kitchen supplies.

Keywords: sustainable cooking, recipe adjustment, ingredient substitution

This report presents the development of a mobile expense-sharing application built with React Native and Expo that addresses the limitations of existing solutions like Splitwise. The app enables users to split bills, track shared expenses, and manage group finances without the restrictions commonly found in commercial alternatives, including expense limits, advertisements, and paywalled features. Key functionalities include real-time balance tracking, group expense management, automatic equal splitting calculations, and comprehensive expense history visualization. The implementation utilizes Clerk for user authentication, AsyncStorage for local data persistence, and Expo Router for navigation, while the backend handles expense calculations and multi-user data synchronization. User testing revealed that the clean, mobile-first interface significantly improves the expense tracking experience compared to existing solutions. The application successfully processes unlimited transactions across multiple expense groups while maintaining accurate balance calculations and providing instant data refresh capabilities. Performance testing demonstrated reliable real-time synchronization between users in shared groups, with consistent data integrity across all expense splitting operations. The final application provides a complete, advertisement-free alternative to commercial expense-sharing platforms, offering all essential features without premium restrictions. This project demonstrates the feasibility of creating user-focused financial applications that prioritize functionality and accessibility over monetization strategies, resulting in a tool that effectively simplifies shared expense management for roommates, travelers, and friend groups.

Keywords: Expense Management, Mobile Development, Data

This project presents the design and implementation of an educational game that utilizes virtual reality to help students practice elementary level math concepts. This game was developed in the Unity game engine utilizing OpenXR standards to ensure broad compatibility with all industry-standard virtual reality headsets. Development of this project included the refinement of a VR control system and C# scripting of robust, interactive physics objects utilizing 3D models designed in-house. Testing for this game was conducted with both an XR device simulator ported into the game engine and Meta Quest 2 VR headsets. The immersive and tactile nature of this educational game offers an advantage over similar games through the effective utilization of VR technologies.

Keywords: Virtual Reality, Education, Unity, C#, Blender, Game-based Learning, Math Education

Skribbled is a modern collaborative note-taking platform created to support students, educators, and professionals in capturing and organizing their thoughts with ease. Designed for both individual use and group collaboration, Skribbled offers a clean, intuitive interface that makes writing, formatting, and sharing notes simple and efficient. Users can write in rich text, incorporate mathematical expressions, and structure their content with Markdown—all within a smooth real-time editing environment.

Whether you're working on class notes, project documentation, or brainstorming ideas with others, Skribbled provides a flexible space that adapts to your workflow. Fully responsive across desktop and mobile devices, it ensures users can access and edit their content anytime, from anywhere. By combining clarity, collaboration, and convenience, Skribbled helps transform note-taking into a more productive and enjoyable experience.

Keywords: Collaboration, Note-taking, Real-time

Sketch Scape is a web-based multiplayer drawing and guessing game built with React.js and Firebase. Players take turns drawing words while others try to guess them in real time. Key features include a drawing tool, and a Party Mode for private games with passwords. We used Firebase to manage game state, user login, and real-time updates. During development, we focused on fixing bugs, improving the user interface, and keeping everything in sync for all players. Final steps included user testing, performance tuning, and preparing all project materials. The game shows how cloud tools and web technologies can be used to build a fun and interactive experience.

Keywords: multiplayer web game, real-time interaction, Firebase, React.js, drawing game, user authentication, game state synchronization, UI design, theme-based gameplay, collaborative gaming