About Me

Hey! My name is Dennis Koshta, I'm a sophomore at UC Santa Cruz, where I'm studying to get my B.S. in Computer Science (with a minor in Statistics). I've been interested in AI and programming since my childhood, and so I've decided to create a website portfolio where I'll be posting my projects, research, resume, and other relevant content. You can connect with and learn more about me by going to my LinkedIn, but this website should have pretty much the same info, ith the addition of the project pages and their source code (in the github repo), which I welcome you to explore! Enjoy.

My Projects

A showcase of various projects that I've worked on.

Hand Motion Tracking

Utilizing MediaPipe NodeJS packages, this project provides real-time tracking and graphical visualization of hands and fingers by processing the input from a webcam. The implementation leverages the powerful machine learning models and algorithms provided by the MediaPipe library to accurately identify and track the positions of individual hand landmarks, enabling efficient and precise representation of hand gestures and movements in a visual form. This technology can be applied in a variety of domains, such as computer vision, human-computer interaction, and virtual reality, among others.

Six Degrees of Kevin Bacon

This C++ program computes the shortest path between two actors in terms of degrees of separation using their movie appearances as connections. Takes a formatted input text file containing actor pairs and outputs the corresponding movie/actor connections in a separate output file. Optimized C++ code using an adjancency list data structure and BFS to find the shortest path efficiently. Click on this project tile to download a zip file with the source code and instructions on use. (Please reach out to me for the archive password, as I cannot share it publicly. This is due to the code being for a university class; I must verify you're not a student before granting access to the assignment solutions.)

Art Canvas

The application presents a user-friendly digital drawing platform that enables users to effortlessly design and produce a wide range of simple artistic creations. It incorporates features such as adjustable stroke thickness, the ability to stamp a fractal tree pattern, extensive color customization options, and dedicated buttons to save the artwork or clear the entire canvas. This artistic tool uses the JavaScript programming language and the HTML5 Canvas element, resulting in a seamless and responsive user experience.

The Monkey's Paw

This project harnesses the power of OpenAI's GPT API to grant users their wishes, with a twist! Inspired by the famous short story "The Monkey's Paw" by W.W. Jacobs, this project aims to provide a fun and engaging experience for users. The project is built using Python and the OpenAI GPT API for the main functionality, and leverages Flask for the web application backend. To ensure the security and privacy of the API key, environment variables and google-cloud-secret-manager are used to keep it secret even though the code is hosted in a git repository.

Multithreaded HTTP Server

This project uses C to implement a simple multi-threaded HTTP server that serves multiple clients simultaneously using a thread-safe queue. The server ensures that its responses conform to a coherent and atomic linearization of the client requests and creates an audit log to identify the linearization. Uses the Pthreads library and supports GET/PUT methods. Click on this project tile to download a zip file with the source code and instructions on use. (Please reach out to me for the archive password, as I cannot share it publicly. This is due to the code being for a university class; I must verify you're not a student before granting access to the assignment solutions.)

Phoneme Recognition from sEMG Signals

This is a Github repository for the Machine Learning and Data team of the NeuroTechSC organization. The main goal of this project is to detect subvocal phonemes from surface Electromyography (sEMG) data using a Deep Neural Network model implemented in Python. The linked repository provides the Python notebooks for the sEMG data preprocessing and various model trainings, as well as data and additonal information/files necessary for context. Currently in the process of expanding to more phoneme classes, currently achieving ~100% test accuracy on 5 phonemes + silence. Uses Pandas for data proceesing and Tensorflow/Keras to implement an RNN/LSTM model.

My Resume

This is a one-page summary of the experience, skills, and certifications that I have acquired in my time learning computer science. It is only an overview, and omits some of the less relevant work I have done, so if you want to learn more about my projects or experiences, please don't hesitate to contact me. You can also access the pdf at this link.

Contact Me

I'm currently looking for a job, so if you have any questions about my projects, or would like to discuss a job offer with me, please feel free to send me an email or contact me using the form below.