2026 Computer Science Capstone Symposium

The Computer Science Department senior capstone presentations will take place Friday, May 1st and Saturday, May 2nd in Rieke Science Center Room 103A on the Pacific Lutheran University campus.  If you’d like to join the capstone virtual session, please email Associate Professor Renzhi Cao at caora@plu.edu.

Friday, May 1st

1:45pm – Creative Hub 
Seven Son (BS), Davine Mungai (BS)

Many portfolio platforms require users to pay in order to maintain the quality of their uploaded images. The rise of generative AI has introduced new concerns around data privacy and ownership. Some platforms have been criticized for using user-generated content to train AI models, often without transparent consent mechanisms. In addition, opt-out processes are frequently unclear or difficult to navigate. Creative Hub addresses these issues by providing an invitation-only portfolio platform designed specifically for freelance creative professionals. Creative Hub is built using Next.js and Supabase, with PostgreSQL row-level security (RLS) to protect user data and ensure that only authorized users can access their information. Creative Hub was created to provide a secure platform where creatives can freely share their work and receive commission requests, all within a single, trusted, AI-free environment.

2:15pm – Campus Life Companion
Kira Mettler (BS)

Campus Life Companion is a cloud-based tool designed to make Resident Assistants’ and Community Directors’ jobs easier by automating one of the most time-consuming tasks: scheduling. Built with JavaScript and integrated into Google Workspace, the application uses a custom algorithm to process RA availability and CD preferences, automatically creating fair and accurate rosters for Desk and Duty shifts. The app runs entirely online, using Google Sheets as the interface with optional Google Calendar automation. Through multiple rounds of testing, the system proved highly effective and has been shown to reduce scheduling time by over 80%, turning up to six hours of manual work into just a 30-minute task. Ultimately, this tool handles the heavy lifting, allowing Campus Life staff to spend more time on what really matters: building community and supporting students.

2:45-3:05pm – Break

3:05pm – SLUMBER PARTY
Reina Buen (BA)

“SLUMBER PARTY!” is a multiplayer platform-fighter video game, inspired by titles like Super Smash Bros. and the Platform Fighter Engine. It combines creative design with core gameplay systems, focusing on combat, animation, and online multiplayer functionality. Rather than delivering a fully polished experience, this stage focuses on establishing a strong foundation for the game. Developed in Unity using Netcode for GameObjects and Unity Relay, the project prioritizes scalability and long-term growth. By building flexible systems and a clear architecture, it is designed to support future expansion and iterative refinement.

3:35pm – Differential – Moment.dev (Partnership)
Gabe Sanders (BS), Marvin Wocheslander (BS)

When you use an app, the screen is supposed to update instantly, but the main computers behind the scenes often struggle to send the new information fast enough to keep up. To fix this, we teamed up with a startup called Moment.dev and their CEO, Alex, to build a clever tool called “differential.” Instead of making the computers send a huge pile of data over and over again, our tool only sends the exact little pieces of information that just changed. It uses some smart math behind the scenes so that even if lots of people are editing the same thing at the exact same time, nobody’s work gets mixed up. Moment.dev is already using it in real life to make their apps run super fast and keep everyone’s screens perfectly in sync!

Saturday, May 2nd

9:30am – Compacts: Foundations for an Adaptive Personal Scheduling System
Andrew Park Lee (BS)

Compacts is a personal scheduling system designed to help users turn tasks, availability, and real-world time constraints into a usable daily plan. The project focuses on building the foundational scheduling architecture needed for realistic time planning, including task management, time tracking, availability modeling, schedule generation, and recovery when plans change. The current version emphasizes a deterministic and constraint-aware planning workflow, while the long-term goal is to extend this foundation into a more adaptive, data-driven scheduling system that improves schedule realism over time.

10:00am – Custom Space Rendering and Procedural Terrain Generation in a Minecraft Modification
Kade Levien (BS)

This project is a modification of the block-based video game Minecraft, colloquially called a mod, and is composed of two distinct parts. The first is the procedural generation of terrain, where I use different types of noise algorithms to create unique looking shapes of the land. The second part of the project is about the custom rendering of planets and space, with orbit logic, lighting, and textures. Each rendered planet represents a unique terrain generation algorithm. When the player gets close enough to a planet while in space, they are teleported to the surface of that planet. To modify the game and register the Java code with Minecraft, the NeoForge modding API was used. This project was inspired by other video games in the space exploration genre, with the goal of bringing some of this exploration to Minecraft.

10:30am – RoomStudio
Cody Nutter (BS), Gavin Bonham (BS), Randy Nguyen (BS)

Finding the right furniture in the right size is inconvenient, especially in the age of digital shopping, often requiring extensive research and the risk of having to return items that aren’t the right size. RoomStudio is a 2D multi-platform desktop application built to help homeowners and renters plan and furnish their living spaces. Users can design layouts with precise, real-world measurements to make managing any space easy. RoomStudio enables users to outline their space through the custom placement of walls, doors, and windows. Users can furnish their rooms using a built-in furniture library, either by entering custom dimensions or importing items directly from an integrated Amazon catalog. This gives users a near-endless canvas to view their existing furniture, while exploring how new furniture can fit in their rooms. RoomStudio reduces the risk and guesswork of furniture selection, making the process of furnishing any space straightforward and stress-free.

11:00-11:20am – Break

11:20am – Real Time GPU-Accelerated Rasterizer and Ray Tracer
Keaton Calvert (BS), Shiori Nakayama (BA)

This project is an exploration of GPU-accelerated real time ray tracing, and how it compares to traditional methods. This consists of an OpenGL real-time renderer that includes a traditional rasterization pipeline, as well as a real-time ray tracer written in a compute shader. These two components can be switched between at will, and load in the exact same scene data, allowing a one-to-one comparison of the two algorithms.

A primary emphasis is placed on reflection and refraction, as those are often the characteristics that visually set the two algorithms apart. Effects like these are much more realistic in a ray tracer, but much slower. With this, the differences between the two can become clear. The framerate of the two algorithms are compared as polygon count is increased. Also, the visual effectiveness of the high speed lighting, reflection and refraction approximations will be tested in the rasterizer. With these comparisons, this project provides a strong educational and visual demonstration of the differences of the two algorithms in a real-time context.

11:50am – Jerry’s Meandering Game
Meghan Buchanan (BS), Jonathan Ochoa (BS), David Luna Guitron (BS)

In the United States some representation is more equal than others. Gerrymandering refers to the manipulation of electoral district boundaries in order for political parties to maximize their political advantage in the House of Representatives. With the help of Jerry and his Meandering Game, we aim to educate the public about this practice through Gerrymandering puzzles. This web-based game was implemented with Phaser + React, and includes includes a dedicated story mode, an infinitely growing map mode, and a less explored multiplayer mode in which you can compete against your political enemies or an AI built on search algorithms. To evaluate all completed puzzles, a statistical ensemble generation technique known as Markov Chain Monte Carlo (MCMC) was implemented using Python and the gerrychain library. MCMC has been increasingly explored with real-world redistricting data, and serves as an efficient way to generate a baseline distribution of what we may expect to see based on the geography of a map. This project combines our collective interest in mathematics, computer science, and political advocacy into an engaging, educational experience.

12:20-1:10pm – Lunch

1:10pm – Tatari AI – College Advisor
Abel Legesse (BS), Biniam Hailu (BS), Yonatan Bayeh (BS)

Navigating college course requirements can be overwhelming. Students often end up taking courses that do not count towards their major, missing requirements they did not know existed, or feeling lost when trying to plan their path to graduation. Tatari AI is a web app that gives every PLU student a 24/7 academic advisor, right at their fingertips.

The system reads PLU’s course catalog and each student’s CAPP report. When a student asks a question, the system searches the catalog using both semantic and keyword matching to find the most relevant information before generating a response. Students can also view their degree plan as an interactive graph that shows prerequisite chains and which courses they have already completed. Major and Minor recommendations are surfaced based on each student’s current standing, and all progress is saved across sessions.

Tatari AI was built with Next.js and React on the frontend, FastAPI for the AI service, and AWS for infrastructure. It uses GPT-4.1-mini with a multi-step pipeline that classifies the student’s intent before generating an answer, reducing the chance of an inaccurate response.

1:40pm – Gear Fitness: A Social Fitness App for Tracking Workouts and Building Community
Bryant Truong (BS), Kobe Cortez (BS), Alton Matovu (BA), Max Lopez (BS)

Staying consistent at the gym is harder than it needs to be. Lifters waste time digging through old logs to find last week’s numbers, struggle to know when they’re ready to progress in weight or reps, and lack a clear view of their progress over time. Motivation often fades without a sense of community or accountability. 

Gear Fitness is a mobile application built with React Native and Expo that addresses these pain points by combining workout tracking with a social feed, enabling users to log their training while staying connected to friends and the broader lifting community. Core features include Google OAuth authentication, detailed workout logging, reusable routines, progress history, streaks and an AI workout assistant. A social layer lets users post completed workouts, follow others, like and comment on posts, and receive push notifications for engagement. The Spring Boot backend persists data to a PostgreSQL database and is deployed on AWS.