Skip to content
CS Capstone Handbook

Project Showcase

We list some noteworthy past projects. More projects will be added soon!

NextJudge (2023-2024)

New Product Web

NextJudge is a suite of services, tools, and applications for creating programming platforms (framework), facilitating programming contests (cli tool), and showcasing programming prowess (platform). The primary product is the NextJudge platform, built with competitive programming in mind.

The NextJudge toolchain comprises a secure code-execution engine, an API gateway, a web application, a CLI tool, and a data layer, all of which are 100% self-hostable.

The Unknown Venome (2023-2024)

Consultancy Web Visualization

The Unknown Venome is a platform that stores, visualizes, and analyzes the Venom Biochemistry Lab’s Proteins at Oregon State University. Search and filter hundreds of venom proteins, view their 3D structure and computed algorithms, and compare with similar proteins such as Foldseek or TM-Align.

Immersive Animal Anatomy Visualizer (2023-2024)

Consultancy AR/VR Visualization

Helping Veterinary students learn anatomy through an environment that allows for immersive, collaborative studying in VR.

Studying anatomy in VR offers an interactive, hands-on experience that allows users to engage with models in a dynamic way, unlike static textbook learning. In this virtual environment, users can explore, move, and manipulate anatomical structures, making it easier to understand complex details through direct interaction.

The Immersive Animal Anatomy Visualizer is distributed via the Oculus App Lab, this means updates/changes can be easily downloaded by users like any other applications on the Oculus Quest 2.

Spatiotemporal Temperature Fusion Network & StarFM (2023-2024)

Research Simulation

Using machine learning to turn unreadable or missing satellite images into usable and rich data.

The Spatiotemporal Temperature Fusion Network (STTFN) is a multiscale convolutional neural network used for modeling nonlinear relationships in Land Surface Temperature (LST) satellite imagery. It predicts missing or damaged data by using two convolutional networks to generate forward and backward sequences, inferring a “middle” sequence. The model leverages MODIS (low-resolution) and Landsat (high-resolution) satellite data to estimate high-quality LST images. The implementation, trained on Oregon State University’s HPC Clusters, uses metrics like RMSE and SSIM for performance and includes a comparison with the STARFM model.