Marius Shepherd

To explore and learn more about cybersecuirty, NIDS, and packet analysis, I created an NIDS from scratch with t-shark, usable as a desktop application. I wanted this to be a learning experience for myself, and the users, so I focused on creating a UI that easily explains NIDS concepts and functionalities to the user, and how packet analysis works.

This helped me do proper research on what a blue team member may do in these secuirty events and also learn more about how network intrusion happens from the lense of packet analysis. I am planning to expand the current application to work for not only WiFI network interface based packets but other network interfaces on MacOS!!

Recall-App is an easy to use mobile application made to enhance active recall among dementia patients. Users can register people based of their faces by taking a photo through the React Native application, and sending Base64 image data to the back-end to store in a SQLite database. Then, patients can scan individuals faces to recall what their names were, which enhances active recall helping with memory loss. The API was created through Flask, and I personally deployed an OpenCV image similarity program that compares the uploaded image with registered images in the database by first cropping the face in each image with Image Cascade, and comparing them with mean square error. If the mean square error is beneath a threshhold, it outputs that registerd image's name. Although simple, this provides a streamlined and fast response to API calls in sending whole images. This project was presented to judges at the closing ceremony. This was my first experience in back-end developement with Python and OpenCV, and explored my interest in computer vision.

A full-stack web application making use of a Clarifai web API, React, Express and PostgreSQL to detect explicit content within user uploaded image. The website has user authentication and user registration, which is done through appropiate API call to: update the rank of the user, which is determined by the number of image uploads of the user (PUT), register the user (POST) and autheticate the user (GET). This project was a culmination of my knowledge in web developement, and I seek to further expand the website.

An interactive website using Tensorflow.js, HTML, CSS and Javascript to learn and use the Japan Sign Language Alphabet. The website utilizes Teachable Machine to detect sign language characters in real-time by sending each frame of a webcam feed to the machine learning model I created with Teachable Machine. The result is displayed as a percentage of the prevalence of each character in the feed in real-time. I also utilized async and await features in ECMAScript to properly handle promises to minimize load times in an already heavy application. This was my first web application I created from scratch and instilled fundamentals of HTML and CSS, along with Javascript DOM manipulation.

A Unity top-down RPG game, created with C# and the help of Michael Doyon. This was an opportunity to explore my interest in game developement and learn developement fundamentals with OOP princicples and top-down models in software developement. I wish to further expand this game and incorporate more complex gameplay mechanics in the future.