Skills
Education
Johns Hopkins University
Work History
Johns Hopkins Applied Physics Lab
Innovative Defense Technologies
Jataware
General
I have multiple years of professional software engineering experience, as well as experience working with many state-of-the-art machine learning techniques.
At Jataware, I worked on a variety of research and technical projects, with focuses on machine learning, and data science applications. For example, I developed a Archytas, a library for using large language models to perform task via the Reason and Action (ReAct) method. There's also the related Boxytas project I authored, for performing Retrieval Augmented Generation (RAG) as well as identifying causal relations between topics, grounded over a corpus of PDF documents. I also researched and implemented approaches to Selective Harvesting (i.e. efficient search + filtering over enormous graph networks), implemented a highly optimized graph-neural-network convolution algorithm in CUDA, and also worked on computer vision approaches to align paper maps with digital coordinates for the USGS georeference challenge. Plus lots of other interesting machine learning applications and research here and there.
At IDT, I focused on both machine learning, and front end development. Of note, I designed a novel machine learning architecture for efficiently allocating compute resources to minimize the execution time of High Level Architecture (HLA) federated simulations. I also implemented a custom time series anomaly detection ensemble model in Julia, and developed the React UI for visualizing the results.
During my master's coursework I implemented a variety of ML algorithms from scratch, including MLP, SVM, Expectation Maximization, PCA, autoencoding, and autocorrelation/cross-correlation. Additionally, I've worked with CNNs, VGG, ResNet, U-net, Viola Jones, and a variety of other architectures. For a capstone project I developed a novel architecture that leveraged the WaveNet vocoder model paired with a custom convolutional transformer network to create a realistic choral voice synthesizer. Prior to that, at an internship with the Johns Hopkins Applied Physics Lab (JHUAPL), I worked on machine learning capabilities for controlling a robotic limb by analyzing electromyogram (EMG) signals in an individual's upper arm.
For my undergraduate coursework, I mainly focused on mechanical design, and pure software development. Additionally, I participated in the JHU Robotics Club, where I worked on several interesting robotics projects, including a picture drawing robot arm, a defintely not beer pong robot, and a Balancing tour guide robot. I also was a member of the JHU Rocketry Club, where I earned my level 1 High Power Rocketry certification, attempted unsuccessfully to earn my level 2 certification, and participated in the 2018 Spaceport America Cup. And lastly, whenever I had the opportunity, I enjoyed participating in the JHU hackathons held twice a year, leading to projects like Ensemble (Hacking Harmony), Boat Simulator, and Bueller Board.
In my spare time, I've worked on a number of interesting side projects, including a custom programming language, a deep learning music synthesizer, several video games, as well as a few other odds and ends. Typically, I like to work on things that are at the intersection of machine learning, music, or game development, but I also frequently find myself working on completely unrelated things, e.g. hydroponics, sewing, or celestial navigation.