Publications

• • The Role of Privacy Guarantees in Voluntary Donation of Private Data for Altruistic Goals
  • Ruizhe Wang, Roberta De Viti, Aarushi Dubey, Elissa M. Redmiles.
  • Under Review. arXiv preprint arXiv:2407.03451, July 2024.
  • click for preprint
• • Crowd Microservices Hackathon: Utilizing Crowdsourcing for Microtask Programming on a Large-Scale
  • Aarushi Dubey, Thomas LaToza, Emad Aghayi.
  • George Mason University Journal of Student-Scientists' Research, November 2019.
  • click for abstract
• • Using Machine Learning to Predict High School Student Employability – A Case Study
  • Aarushi Dubey, Muthukumara Mani.
  • IEEE International Conference on Data Science and Advanced Analytics (DSAA), October 2019.
  • click for poster paper

Projects

• • Stenographic Plugins for Vulnerable Audiences
  • Developing a messaging application focusing on balancing in-group user autonomy and situational relevance by implementing group/channel-dependent user-choice pseudonymity; big-group random identity swapping; in-message stenography; and geo-fencing turn on/off abilities.
  • click for writeup
• • PrivaZy
  • Curated for a privacy-awareness centered movement to provide insight and conversation surrounding Generation Z’s mental models on social and technical implications of digital privacy.
  • click for information
• • Differential Privacy and Reconstruction Attacks
  • Worked with UMD faculty to develop and empirically test a variety of reconstruction attacks on anonymized summaries of unprotected data. Synthetic data was developed through the Harvard OpenDP library for differential privacy, and reconstructed through the optimization library Gurobi. Focused on examining the role of the epsilon parameter in differential privacy.
• • Exploring Pedagogical Approaches and Student Learning Strategies to Computer Science in Higher Education
  • Synthesized examination of undergraduate students’ understanding/comprehension & learning preferences of computer science concepts + their alignment with the science of learning through a multi-method qualitative study with Dr. Samantha-Kaye Johnston at the University of Oxford.
• • Interactions Between Prior Exposure to Digital Technology and Cultural Perceptions of Technology
  • Wrote ethnography examining how do (prior exposure to digital technology concepts) and (a student’s cultural background & cultural perceptions of technology) interact with one another to influence a student’s learning skill sets and comfortability with technology in college through the use of semi-structured ethnographic interviews
  • click for writeup
• • Technology Privacy Concerns of Students
  • Worked with others to perform a case study on college students’ interpretation of privacy and their related concerns amidst the technology shifts of the private and public spheres in the COVID-19 endemic.
  • available upon request
• • Student Attitudes Towards CS [Education ∪ Industry ∪ Biases ∪ Politics]
  • Worked with Dr. Jose Calderon at the University of Maryland at College Park to examine computer science student perceptions and opinions on various social spheres CS students interact with, varying in education, industry, biases, and/or politics.
• • Climate Change Prediction-to-Action Pipeline
  • Developed a deep learning pipeline that combines RNNs and GPT-based reccommender systems to 1) predict on what climate non-actions to a region will do, and then 2) provide policy actions in a palatable form for non-science experts.
  • click for writeup
• • Deep Learning Varieties
  • Built deep learning models on image segmentation, text generation, classification, and more with PyTorch and TensorFlow & Keras - developed variational autoencoders [VAE], recurrent neural networks [RNN], convolutional neural networks [CNN (U-Net)]; and built mini GPT models.
• • Factors in Education Inequality
  • Developed with others a rudimentary data science pipeline tutorial aiming to confirm if education inequality in the USA is reflected by national math and reading examination differences, and predict how future years of education inequality will be if the current USA education system/institution is maintained; all throughout utilizing machine learning techniques and statistical analyses.
  • click for tutorial
• • DevSecOps
  • Developed Mulesoft APIs and RESTful Web Services that interacted with Azure Data Lake Storage systems and AWS S3 systems for data integration through Mulesoft Anypoint Studio; transformations in JSON, XML. Projects built through Maven, deployed through Jenkins pipelines, tested through Postman, and pushed to Bonobo Git servers with command-line Git. Additional projects containerized with Docker and depoyed onto Kubernetes clusters through Rancher + AWS EC2 instances. Aided in setting up authorization framework OAuth 2.0 and identity layer OpenID Connect with identity provider Okta to secure Mulesoft APIs on the Anypoint Platform.