Aidan Stoner
computer science student at The College of New Jersey
specializing in data science
with a minor in mathematics
experience
Research Intern
Princeton Plasma Physics Laboratory (PPPL)
June 2026 to Present
Conducted research on AI and information retrieval systems applied to fusion and plasma domains. Developed a document retrieval framework, accelerating the integration of retrieval systems into applications. Evaluated accuracy of multiple retrieval systems, finding proposed system achieves 8% higher accuracy than baseline.
Software Development Intern
New Jersey Department of Military Affairs
June 2024 to Present
Led the adoption of a modern web framework, replacing legacy technology through research, collaboration, and advocacy. Directed the full-stack development of three web applications, delivering more productive and secure workflows for end users. Utilized ASP.NET (Core MVC, Webforms) with C#, SQL, JavaScript, JQuery, and Bootstrap.
Computer Science Tutor
TCNJ Department of Computer Science
August 2025 to Present
Providing tutoring in computer science, focusing on programming, algorithms, and problem-solving to help students improve their understanding and excel academically.
Quantum Computing Researcher
TCNJ Department of Computer Science
August 2025 to May 2026
Conducting undergraduate research to build a foundation of quantum computing, with an eventual focus on validation of SHA-256 uniformity.
projects
Document Retrieval Framework
Document retrieval, abstracted.
A Python framework that provides a common interface for the document retrieval process. Working across many different retrieval systems, each implementation required bespoke tooling despite following the same underlying pattern. This framework abstracts those implementations into four components, each responsible for a distinct stage of the pipeline.
started June 2026
The Transformer Architecture, from scratch.
This model is a complete implementation of the state-of-the-art Transformer architecture, built using only basic PyTorch layers and tools. Given a sequence of tokens, this model excels at predicting the next token of the given sequence. This simple task of predicting the next word is the foundation of how modern LLMs are able to generate responses that are not only human-like, but also relevant to the prompt it was given.
started January 2026
Explore academic papers through interactive, stellar graph visualizations.
Most academic databases provide flat lists of search results, making it difficult to understand the cross-disciplinary relationships and foundational literature within a field. Stellar Papers reimagines the literature review process by turning a database of 200,000+ research papers into an interactive, physics-simulated citation graph. Its focus on highly optimized data pipelines and fast rendering provides an immersive and openly experimental environment for discovering the connections that drive academic progress.
started February 2026
Predicting levels of Alzheimers using a CNN with MRI images.
A convolutional neural network (CNN) built in PyTorch to classify the severity of Alzheimer's disease from MRI brain scans. Trained on the MRI Scans Alzheimer Detection Dataset via Hugging Face, the model achieves a best test accuracy of ~99% across four severity classes.
started December 2025
and more... visit my github
certifications
