Evan Wushke

Computer Vision Schematic Parser

Made with Python, YOLO, U-Net, and TrOCR. A multi-stage computer vision pipeline that converts hand-drawn electrical schematics into structured XML. A fine-tuned YOLO model detects schematic symbols and text regions, a U-Net segments the wire topology, and TrOCR reads handwritten component labels. The pipeline then reconstructs the full schematic by linking wires to components and exporting a clean digital representation.

Mandlebrot Convergence Visualizer

Made with C, JavaScript, HTML, and CSS. This web application visualizes the convergence of points in the Mandelbrot set, allowing users to explore the fractal by panning and zooming. It renders escape-time iterations in real time to reveal the intricate boundary structure of the set, offering an interactive way to study how complex numbers behave under iteration.

Finance Research Dashboard

Made with Python, Shiny for Python, pandas, numpy, matplotlib, and Docker. My senior project is an interactive finance research dashboard that pulls live market data from Yahoo Finance and presents stock analysis through a modular tabbed interface, including a data table view, standard deviation and volatility analysis, and Sharpe ratio analysis. The app is built around Shiny's reactive UI/server separation, containerized with Docker and docker-compose for reproducible environments, and packaged as a deployable image ready to be hosted on AWS through ECR and ECS/Fargate or EC2.

Observing the Baldwin Effect in Pac-Man

Made with Python, PyGAD, scikit-learn, and numpy. A research project that combines reinforcement learning and a genetic algorithm to observe a phenomenon in evolution known as the "baldwin effect." Pac-Man agents evolve innate Q-learning weight vectors across generations in a grid-world environment. Over time, the genetic algorithm independently converges on the same weight configurations that in-lifetime learning would have found — demonstrating how learning capacity can guide evolution without direct inheritance.

Disk Scheduling Algorithm Visualizer

Made with JavaScript and Tailwind CSS This web application visualizes various disk scheduling algorithms, including FCFS, SSTF, SCAN, C-SCAN, LOOK, and C-LOOK, to demonstrate how they work and how they compare in terms of efficiency for different data requests. This project was created to help understand important topics in operating systems.

3D Pathfinder

Made with Python, Matplotlib, and Haskell. This project is a representation of the Traveling Salesman Problem (TSP) in 3D Cartesian space. It features a live recursive backtracking visualization and leverages functional programming using Haskell. The purpose of this project was to solve the same problem using both a declarative and an imperative language, providing insights into the differences between programming paradigms.

Formula 1 Dashboard

Made with JavaScript, Tailwind CSS, Node.js, and Express. The F1 Dashboard is a single-page web application that leverages DOM manipulation and asynchronous JavaScript to display data from the Formula 1 API, and the API itself was also re-created as part of this project.

Pacmans Revenge

Made with C and Assembly. Pacmans Revenge is an arcade style variant of the arcade game "Pacman" created for Computing Machinery II at Mount Royal University. This game is intended for the Steem emulator. Almost all reliance on the operating system has been removed, with the exception of the Super() system call. This project had the purpose of learning low-level programming and hardware interactions, as such this project is specific to the MC68000 architecture.

Hi there! Welcome to my portfolio.

I've learned a thing or two from building so many projects.

projects.