蓝莓视频

Joe Fielding & Kyle Richards

"Virtual Reality Puzzle Game"

Client: Dr. Hoang Bui

Our client wants a virtual reality puzzle game application to play on the Meta Quest headset. Our client enjoys solving jigsaw puzzles, but it is impractical for him to repeatedly buy new puzzles. He wants to instead generate new puzzles quickly and then solve them on his Meta Quest headset. Our VR puzzle game application will allow users to solve puzzles in the VR world generated out of pictures of real-life objects. Users can choose both the shape and the number of pieces. Additionally, users will be able to move and connect pieces using real hand motions. There will also be a leaderboard associated with each puzzle so users can see how their solve times compare to others. Finally, there will be a feature to import a picture of a real jigsaw puzzle with its pieces spread out and solve it in the VR world.

Nyrique' Butler & Chase Garnett

Our client is Jordan Dahl, who is the CEO of Points of Control. Points of Control is a course management platform that offers coaching and consulting for entrepreneurs. Jordan is a friend of a friend of Chase鈥檚, and we chose Jordan because when we met with him he presented a clear need for his business and he let us know his availability for feedback and content provision.
The objective of our capstone project is to build a course management web application that focuses on centralizing course access and client progress tracking under a minimal viable product strategy. Jordan explained to us that the company鈥檚 current process was scattered and partially manual which was managed by a 鈥減atchwork鈥� of tools to sell their courses and host content. The pain points we are specifically looking to resolve are the time sinks that are caused by the burden of managing more and more clients overtime, the inconsistency of user experience or having no unified view of client progress, and the lack of visibility to track who has completed what courses.

Brendan Bokino & Nishant Gurung

"PlantWatch"

Client: Dr. Mren Blohm

Plant Watch is a mobile application aimed to address the growing challenges faced by 蓝莓视频鈥檚 Biology Department, having to visit campus greenhouses to check on the environmental conditions, which takes away valuable time from the faculty and staff. The app uses computer vision to let users scan plants and get back species identification, watering schedules, and nutritional deficiency detection, while a networking layer pulls live data from IoT sensors like AcuRite weather monitors and soil moisture probes; providing Dr. Blohm鈥檚 team with important information they need to make decisions on the management of the greenhouses and to set automated rules for device control like triggering sprinklers when soil moisture drops below a threshold. The project aims to deliver something that gives Dr. Blohm鈥檚 team a reliable way to keep the plants healthy and save time. 

Stuart Belvin & Fabrizio Guzzo

"A-Real-Time Hockey Analytics and Coaching Dashboard"

Client: Matthew Barrow, Hockey Coach and Analytics Lead

This project is a hockey coaching tool that tracks in-game statistics live. This allows hockey coaches to make their game-deciding decisions as informed as possible with accurate statistical data. The statistics are logged by authorized audience members and displayed in a live updating graphical dashboard for the coach to view. The game data is then logged, allowing for the creation of a statistical dashboard showing the team鈥檚 progress over all previous games, performance against specific opponents, and individual team member statistics. The software can automatically generate game reports and areas of improvement to create real insights and enhance team performance.

Chloe Miranda & Jonathan Dargakis

After identifying inefficiencies in the current workflow used at Artificial Axon Labs, our client recognized the need for a unified system to process and analyze high-resolution 3D images of artificial neurons and their myelination. At present, researchers must switch between Python, Jupiter notebooks, Bash scripting, and ImageJ to complete preprocessing, rendering, and analysis tasks, often manually adjusting parameters for each dataset. This fragmented process is time-consuming, inconsistent across team members, and makes collaboration and reproducibility difficult.
The goal of this project is to develop a centralized, user-friendly software application that consolidates the entire image-processing pipeline into one cohesive platform. Researchers will be able to upload datasets, configure analysis parameters, render and view results, and export publication-quality images without navigating multiple tools. The system will also maintain a structured history of processed datasets, allowing users to track configurations, compare outputs, and reuse previous settings. With this application, we aim to aid in streamlining workflows, improving consistency, and supporting reproducibility in their research.

Oscar Roat & Dylan Morales

Client: William Tikiob

This project is an inventory platform for CoolSys, an HVAC contractor. The system will centralize information for jobs, most importantly, lists of materials at each site. Currently, on-site employees do not know what materials are supposed to be delivered. This platform will allow project managers to upload purchase orders to input material lists. On-site employees can view the lists, as well as compare them to delivered materials. Additionally, there will be both a web app for those in the office and a mobile app for those on-site.

Yohann Gouin & Joel Robinson

鈥淒igital Watershed鈥�

Client: Billy Friebele (Program Director of Studio Art)

This project focuses on designing a sustainable, web-based system for organizing and visualizing a complex, multi-modal body of artistic research. The application functions as a 鈥渄igital watershed,鈥� allowing documents, maps, audio, video, 3D scans, and other media to be collected, tagged, and structured into larger conceptual categories. By unifying diverse research materials into an intuitive interface, the project supports long-term research stability, enhances creative workflows, and offers an interactive way to explore how individual research threads flow into broader artistic themes and finished works.

Jordan Lim & Reece Watkins

"AI-Driven Client Retention & Referral Engine"

Client: Rodney Fontil, Hearth Realty

This project is a custom-built digital platform designed to solve one major problem: the fact that most clients forget their agent鈥檚 name after closing. We are building an AI "Concierge" that connects clients to their agent鈥檚 trusted list of local pros鈥攍ike plumbers, lawyers, or even the best neighborhood cafes. By providing this 24/7 service, the agent stays "top of mind" for years, not just during the sale. The goal is to turn a one-time commission into a lifelong relationship, making it a no-brainer for clients to return to their agent for their next move or to refer their friends.

Aidan Marshall & Vilnis Jatnieks

Aidan and Vilnis partnered with the Karson Institute for Race, Peace, & Social Justice to develop the Karson Digital Library. This project addresses measurable pain points: excessive staff time spent on manual inventory logging, inconsistent checkout tracking, and limited patron access to the collection.
They will deliver a full-stack web app with role-based access for administrators, staff, and patrons. Features like ISBN auto-lookup and barcode scanning will cut inventory logging time, while an intuitive interface with personalized book recommendations will make the collection discoverable to the wider community.

Crystal Ajayi & Zoe Willis

"Glory Harbor Works"

Client: Pastor Victor Akinde

Glory Harbor Works is a full-stack web application designed to support a church community by centralizing important functions such as meeting scheduling, sermon streaming, and prayer requests into one accessible platform. The idea for this project comes from firsthand experience serving in the media department at STGCI-Glory Harbor, a growing church under the leadership of Pastor Victor Akinde.
The site is intended to become a central hub for current members and future visitors to gather. It will provide easy access to important information, events, and resources for anyone looking to connect with the church. This project is especially meaningful because it addresses real needs within a community while helping to create a more organized, welcoming, and connected church environment.

Oselunosen Ehi-Douglas & Justin Dorsey

鈥淧opulation Model Calculator (PMC)鈥�

Client: Dr. Suzanne Keilson, Associate Professor

The Population Model Calculator (PMC) is a web-based educational tool designed to help students and instructors explore and analyze mathematical population models in an interactive and intuitive way. The application allows users to compute and visualize models such as exponential and logistic growth, carrying capacity, and differential equation鈥揵ased systems through dynamically generated graphs and simulations. Users can adjust parameters in real time, compare multiple models' side by side, and upload real-world datasets for curve fitting and accuracy analysis. With support for user accounts and instructor-specific features, the platform aims to simplify complex population modeling concepts while providing a flexible and accessible alternative to traditional calculators or specialized software.

Leslie Kim

"Automated Test Generation Using LLM: A Replication Study"

Client: Dr. Henrique Rocha

This project evaluates the effectiveness of Meta's TestGen-LLM method using the Tests4Py benchmark. We assess LLM-generated extended test suites through a set of filtering steps measuring build correctness, flakiness, and coverage improvement.  We compare the overall and filter success rates to those reported in Meta's study.

Hans van Lierop

"Combinatorial Methods for Chip Vulnerability Detection"

Clients: Richard Kuhn, Michael Zuzak, M S Raunak

This research project investigates the effectiveness of combinatorial covering arrays on chip vulnerability detection. There is an increasing percentage of chips being released with significant vulnerabilities, leading to costly respins. Industry standards of randomized testing are no longer suitable for the growing complexity of these chips. Since combinatorial testing has proven successful in software systems, it has become a promising approach to hardware verification.

Nathan Barton

"Optimizing Quantum Circuit Simulators"

Advisors: Dr. David Binkley & Dr. David Hoe

This project explores the optimization of quantum circuits through classical simulations. Quantum circuit simulators use state vector representations; the number of states grows exponentially with the number of qubits, creating a scalability problem. Circuit cutting techniques, such as CutQC, aim to mitigate this problem by partitioning the circuit into smaller subcircuits to be run in parallel. However, there is a tradeoff that the reconstruction stage after parallelization grows exponentially with the number of cuts made. This project aims to analyze that tradeoff and develop a method for determining the number of cuts to make on a given circuit.

Loren Kim

"Evaluating CNNs for Hand Fracture Detection and Classification"

Advisor: Dr. Eric Cui & Dr. Emery Kim

This research project investigates the F1 scores and AUC-ROC curves of ResNet-50, DenseNet-121, and EfficientNet-B3 for classifying hand and wrist fracture detection and classifying fracture type. Physicians may miss fractures on X-rays, especially in the hand and wrist, due to the small anatomy of the numerous bones. Artificial intelligence is exponentially growing in the medical field, and physicians should utilize the highest-scoring model. Previous research has proved that CNNs consistently outperform physicians in X-ray fracture detection. However, little research has been conducted to determine which model outperforms others. The models were specifically chosen as ResNet-50 historically does well on small bone X-rays, DenseNet-121 has been shown to successfully detect subtle texture changes or microfractures, and EfficientNet-B3 utilizes compound scaling, low computation cost, and high accuracy outputs, which are beneficial for the high demand and high speed of diagnosis in emergency rooms.