Matching Items (727)
Filtering by
- Creators: Computer Science and Engineering Program
Description
The objective is to develop a US GAAP and IFRS-based financial reporting software application. The application will reference the relevant codification under US GAAP or IFRS accounting standards and provide the user with a step-by-step guide to recording technical accounting entries dictated under the relevant codification or standard. The software solves the need to hire external expertise to perform such tasks. Through a series of relevant inputs, in-house accounting staff (or students) are afforded the opportunity to methodically work through the accounting topic or arrangement with reference to the codification and accounting standards. The software provides the final deliverables generally expected for financial reporting.
ContributorsMudiam, Nivedh (Author) / Baskaran, Hrishi (Co-author) / Zhang, Helen (Co-author) / Giles, Jackson (Co-author) / Chen, Yinong (Thesis director, Committee member) / Hunt, Neil (Committee member) / Lamoreaux, Phillip (Committee member) / Barrett, The Honors College (Contributor) / School of Mathematical and Statistical Sciences (Contributor) / Economics Program in CLAS (Contributor) / Computer Science and Engineering Program (Contributor)
Created2025-05
Description
This paper aims to introduce a new building block to switching power converter design. A buck converter, boost converter, buck-boost converter, and other topologies utilize a pair (or more) of semiconductor devices that can be abstracted to a power pole primitive. This text explores a method to derive needed design and control models on a traditional power pole. These methods can then be applied to the proposed 5-terminal power pole. Compared to the traditional 4-terminal power pole, the proposed power pole offers greater control over waveform shapes. The potential benefits of the 5-terminal power pole become clear when analyzing the proposed topologies that utilize the 5-terminal building block.
Primarily, this paper aims to provide the knowledge needed to construct (in simulation) and control a novel topology utilizing the 5-terminal power pole. This topology is in many ways similar to a buck converter, so buck converters are studied first in this paper. The analysis process applied to the buck converter is then applied to the proposed converter. By the end of the analysis, the reader should have the information to design these converters. Depending on the use case, the proposed converter has the potential to have reduced output ripple (voltage/current) compared to a buck converter with an equivalently sized passive filter. Therefore, if meeting an output ripple (noise) spec is a design goal, which it very often is, then the proposed topology has the potential to be smaller, in terms of physical footprint, and less expensive, in terms of money, than a buck converter.
Finally, several other uses for the 5-terminal power pole are analyzed. The first use is a current splitter that can be used to make a single input, multiple-output switching power converter. Next, another novel topology is proposed that has two immediately obvious use cases. For one, it can be used to generate an oscillation (with potentially varying frequency or amplitude) while providing significant power, efficiently. Also, it can be used to generate an arbitrary waveform while providing significant power, efficiently.
ContributorsFitzgerald, Zachary (Author) / Kitchen, Jennifer (Thesis director) / Bakkaloglu, Bertan (Committee member) / Barrett, The Honors College (Contributor) / Computer Science and Engineering Program (Contributor) / Electrical Engineering Program (Contributor)
Created2025-05
Description
The integration of Artificial Intelligence (AI) algorithms and tools in cloud computing is revolutionizing business processes with its increased adoption among leading technology companies. By analyzing the AI Cloud solutions that companies offer across diverse industries– including healthcare, gastronomy, streaming applications, and service providers– this thesis determines their impact on overall company productivity and user experience. Some of the platforms offering these tools that were evaluated include those from major technology and business intelligence companies, such as Salesforce’s AI Cloud Tools, Oracle’s Cloud Infrastructure, Amazon Web Services, Microsoft Azure’s OpenAI Service, and Google Cloud. In order to determine the impact of AI Cloud on both company productivity and user experience, this thesis cross-analyzed shifts in metrics such as workload efficiency, customer resource management, sales performance, and financial outcomes following AI Cloud implementation. The initial implementation of AI Cloud can be costly and as an increasingly pervasive technology with potential to attract security threats, it can be met with uncertainty and doubt. Despite these initial disadvantages, the metrics used in this thesis suggest that AI Cloud solutions have an overall positive impact on company productivity and user experience when intentional, proper deployment is exercised. The literature in this thesis suggests that achieving this successful deployment requires consideration of ethical guidelines, security practices, and human-AI integrated mediation when implementing AI solutions. By ensuring that the AI Cloud solutions companies offer are optimal for their processes, have proper safeguards in place for any potential errors and security concerns, and prioritize a positive user experience for both employees and customers, companies can increase their productivity and overall efficiency of their business processes.
ContributorsLecuru, Sophia (Author) / Wang, Amy (Co-author) / Martin, Thomas (Thesis director) / Roumina, Kavous (Committee member) / Barrett, The Honors College (Contributor) / Computer Science and Engineering Program (Contributor) / School of International Letters and Cultures (Contributor)
Created2025-05
Description
The ubiquity of streaming data in real-world applications underscores the critical need for dynamic and adaptive machine-learning models. Streaming data presents the challenge of concept drift, where existing models are no longer effective as new information comes into the model, necessitating continuous model adaptation. Despite advances in continual learning and drift detection, deep learning models remain highly susceptible to catastrophic forgetting when faced with streaming data. This instability hinders their ability to effectively retain previously learned knowledge while adapting to new distributions, leading to significant performance degradation. To tackle these challenges, we propose TemporalGR, a deep generative replay based framework that enables models to learn from temporal streaming data without sacrificing past knowledge. By generating synthetic data from past streams of data, our model maintains stable and efficient performance across evolving distributions. To enable effective generation, LSTM layers and new loss functions are implemented to better model and remember the data. Our experimental results on diverse multi-class temporal streaming datasets showcases some of the potential strengths of the model.
ContributorsLiu, David (Author) / Fu, Yanjie (Thesis director) / Vignesh Malarkkan, Arun (Committee member) / Barrett, The Honors College (Contributor) / Computer Science and Engineering Program (Contributor)
Created2025-05
Description
This thesis presents the design, development, and deployment of an AI-powered chatbot made for Abundant Harvest Aquaponics, an agriculture organization focused on sustainable eco-friendly food production and community engagement. The primary goal of this project was to modernize the organization's customer interaction capabilities through an intelligent, conversational interface that streamlines communication, enhances user experience, and supports operations such as donations, CSA (Community Supported Agriculture) subscriptions, and educational outreach.
Built using natural language processing (NLP) technologies and integrated into the renovated organization’s existing website infrastructure, the chatbot was trained to handle frequently asked questions, guide users through various services, and provide real-time information about aquaponics systems. The implementation process involved iterative design, feedback-based refinement, and aligning with the organization’s mission and audience, which includes health-conscious consumers and donors aged 24–60.
The chatbot has been successfully deployed, but metrics have not yet been collected. Early observations suggest it is functioning as intended, with the potential to improve user engagement, reduce the inquiry load on staff, and enhance overall customer satisfaction as intended. The project is a major stepping-stone for Abundant Harvest Aquaponics. It reflects how AI can help benefit small and medium-sized enterprises and support sustainable, community-oriented agriculture.
ContributorsAlvarado, Ivan (Author) / Miller, Grant (Co-author) / Osburn, Steven (Thesis director) / Hendrix, Charles (Committee member) / Barrett, The Honors College (Contributor) / Computer Science and Engineering Program (Contributor)
Created2025-05
Description
This thesis presents the design, development, and deployment of an AI-powered chatbot made for Abundant Harvest Aquaponics, an agriculture organization focused on sustainable eco-friendly food production and community engagement. The primary goal of this project was to modernize the organization's customer interaction capabilities through an intelligent, conversational interface that streamlines communication, enhances user experience, and supports operations such as donations, CSA (Community Supported Agriculture) subscriptions, and educational outreach.
Built using natural language processing (NLP) technologies and integrated into the renovated organization’s existing website infrastructure, the chatbot was trained to handle frequently asked questions, guide users through various services, and provide real-time information about aquaponics systems. The implementation process involved iterative design, feedback-based refinement, and aligning with the organization’s mission and audience, which includes health-conscious consumers and donors aged 24–60.
The chatbot has been successfully deployed, but metrics have not yet been collected. Early observations suggest it is functioning as intended, with the potential to improve user engagement, reduce the inquiry load on staff, and enhance overall customer satisfaction as intended. The project is a major stepping-stone for Abundant Harvest Aquaponics. It reflects how AI can help benefit small and medium-sized enterprises and support sustainable, community-oriented agriculture.
ContributorsMiller, Grant (Author) / Alvarado, Ivan (Co-author) / Osburn, Steven (Thesis director) / Hendrix, Charles (Committee member) / Barrett, The Honors College (Contributor) / Computer Science and Engineering Program (Contributor) / Department of Military Science (Contributor)
Created2025-05
Description
This paper presents a prototype drone flight controller framework designed for hardware transferability and deep customization, with an implementation focused on the ESP32-S3 microcontroller. Drawing inspiration from existing open-source platforms such as Betaflight and ArduPilot, the proposed framework decouples critical functions into modular software components (for I2C, UART, DShot, and Crossfire communication), each of which can be substituted or extended for new hardware. This modular design eliminates the need to overhaul large swaths of code for nonstandard drone builds, in contrast to the more monolithic structures of conventional firmware. The approach parallels NASA’s Core Flight System (cFS), prioritizing portability and mission-specific flexibility.
ContributorsDavoudi, Keon (Author, Co-author) / Osburn, Steven (Thesis director) / Vuong, Eric (Committee member) / Barrett, The Honors College (Contributor) / Computer Science and Engineering Program (Contributor)
Created2025-05
Description
Email marketing is an invaluable tool to increase engagement by reaching out to potentially interested customers directly (Coursera, 2025). GreenExpectations, a real-estate company based in the greater New Jersey area focused on sustainability, stands to benefit from a web application which would allow them to promote specific sustainability-related products, technologies, or properties to their customer base. The web application “EcoLogin”, built using Next.JS with a Typescript base, provides a solution for GreenExpectations by delivering value to end users with an easy and intuitive method to subscribe to different sustainability-related topics. Additionally, EcoLogin enables GreenExpectation’s administration to export each sustainability topic’s subscribers for effective use in their internal marketing. User information being saved in a deployed PostgreSQL database instance ensures both reliability and security for EcoLogin’s data.
ContributorsTang, Ethan (Author) / Osburn, Steven (Thesis director) / Pokidaylo, Boris (Committee member) / Barrett, The Honors College (Contributor) / Computer Science and Engineering Program (Contributor)
Created2025-05
Description
This project aims to identify particular traits, specifically off-field and non-gameplay, of sports narratives that elevate them to legendary, beloved storylines among the canon of sports history, focusing on the “big three” American sports of baseball, basketball, and football. This was accomplished through an analysis of existing literature on the topic of sports narratives, as well as three case studies of individual narratives from varied sports and points in history. Each study, each representing either a legendary, hate-watched, or forgotten narrative, was broken down into its background, relevant people, events, and contemporary media coverage, and lasting legacy. The various aspects of these studies were then compared and contrasted, with concepts from the literature review being included in the synthesis of the storylines. Ultimately, the presence of a clear protagonist-antagonist dynamic, balanced media coverage, high stakes, and perceived authenticity were determined to be crucial for a sports narrative to gain legendary status. In addition, the notion of authentic coverage was found to have been able to shift public perception of a narrative as well as “resurrect” forgotten storylines of the past.
ContributorsMitchell, Andrew (Author) / O'Flaherty, Katherine (Thesis director) / Boivin, Paola (Committee member) / Barrett, The Honors College (Contributor) / Computer Science and Engineering Program (Contributor)
Created2025-05
Description
This thesis presents the design, development, and strategic decisions of Translatica, a
voice-preserving AI translation platform. In contrast to conventional translation tools
that prioritize textual accuracy at the expense of speaker identity, Translatica seeks
to amplify global communication by preserving the original speaker’s tone, emotion,
and personality throughout the translation pipeline.
The system integrates transcription, context-aware translation, and neural voice
replication to produce emotionally expressive, multilingual video outputs. Originating from a hackathon focused on educational technology, Translatica evolved into a
scalable solution seeking to bring accessible content for everyone. Its primary use
case began in education but now spans creators, enterprises, and institutions seeking
culturally resonant global communication.
This work analyzes current market trends, competitor limitations, and the rising
demand for speech-to-speech translation in education and media. It explores user
interface design decisions, business model strategies, and ethical considerations in
deploying voice synthesis technologies. Reflections from the development process
highlight the importance of clearly defined problem statements and collaborative
iteration in building impactful, user-first AI systems.
This thesis represents one component of a broader collaborative project developed
alongside two other engineers. While this paper focuses on the business strategy,
user interface design, and ethical implications of the platform, the complementary
theses address backend engineering, software infrastructure, and the training and
deployment of AI models powering Translatica’s translation pipeline.
ContributorsHsu, Jeffrey (Author) / Jhaj, Baaz (Co-author) / Ramani, Krishna (Co-author) / Osburn, Steven (Thesis director) / Zhu, Haolin (Committee member) / Barrett, The Honors College (Contributor) / Computer Science and Engineering Program (Contributor)
Created2025-05