Graduate Student Research
Our research-based MSc and PhD degrees provide students with theoretical, practical and analytical expertise, as well as experience in the application of scientific results to real-world problems. Graduate students can pursue their own original research and/or be a part of faculty research projects.
Recently Published
Working alongside faculty members, our graduate students engage in meaningful research and regularly publish their findings as first authors in peer-reviewed scientific journals.
Bakdauren Narbayev
PhD Student, Computer Science
Supervisor: Dr. Mohammad Khalad Hasan
NinjaPort: A Multi-Hand Approach for VR Teleportation
“Having superabilities in Virtual Reality (VR) is not only fun but can also improve user performance. Our work shows that using four virtual copies of the dominant hand enables users to navigate virtual environments more efficiently.”
Moving around in Virtual Reality (VR) can be challenging when the virtual environment is larger than the available physical space. Teleportation is a common solution that allows users to select a destination and instantly relocate there. While newer VR headsets can track hand movements instead of relying on controllers, hand-based teleportation is often less accurate and more physically demanding. To address these challenges, we introduce NinjaPort, a novel hand-based technique that lets users control four virtual copies of their dominant hand, each positioned at a different distance. Users can quickly select a teleportation distance by pinching a specific finger on their non-dominant hand. Our results show that NinjaPort improves both speed and accuracy compared with existing hand-based teleportation techniques while requiring less physical effort.
Check out their paper in 2026 IEEE Conference on Virtual Reality and 3D User Interfaces (VR).
Published April 2026
Paavai Manimaran V
MSc, Computer Science
Supervisor: Dr. Yves Lucet
Optimal network topology and alignment design for minimizing construction costs of wind farm access roads
“An optimization framework is developed for wind farm road design in complex terrains. By integrating construction cost modelling and engineering constraints, the work determines how optimization approaches can reduce cost, improve efficiency and decision making in road design.”
Designing roads for wind farms in hilly or uneven landscapes is complex and often expensive. Engineers must balance safety, environmental impact, and construction costs, and these designs are commonly created manually, which can be time-consuming and inefficient. This work introduces an automated, optimization-based approach to improve how wind farm road networks are planned. The framework uses detailed terrain information to identify road routes that are safer and less costly to build. Rather than relying on trial and error, the method systematically evaluates many possible design options and selects those that minimize earthwork and construction costs. When tested on real wind farm sites, the approach achieved cost reductions of up to 14% compared to the designs used. This work supports better engineering decisions and more efficient development of renewable energy infrastructure.
Check out their paper in Computer-Aided Civil and Infrastructure Engineering.
Published April 2026
Soonuk Kwon
PhD Student, Computer Scienced
Supervisor: Dr. Pourang Irani
Draped Surfaces: A Contour-Adaptive Interface Overlaid on the Physical Environment for Mixed Reality Workspaces
“Our HCI research is grounded in the belief that technology can make people’s lives safer, better support their activities, and become more humane. This study is one step toward that goal, exploring how mixed reality can enable unobtrusive spatial computing that supports our unrestricted activity.”
In spatial computing, current virtual displays often appear as floating flat surfaces, much like conventional monitors. The problem is that, due to their rigid shapes, they cannot be placed easily in environments. As a result, they can obscure the view, reducing users’ awareness of their surroundings and ultimately restricting their freedom of activity. This becomes critical in safety-sensitive settings, such as industrial environments, where workers must remain aware of nearby tools, their hands, and potential hazards.
Our work begins with this concern. To address this question, we explore a new cloth-style display that can be placed over surrounding environments. Our “Draped Surfaces,” an unobtrusive, environment-adaptive virtual display, helps users remain aware of their environment while presenting clear digital content, keeping the physical world more visible while maintaining display legibility. Through controlled studies, our findings show that draped surfaces can better reveal hidden areas while enabling smooth interaction with users’ surroundings.
Published April 2026
Tarik Hasan
MSc Student, Computer Science
Supervisor: Dr. Mohammad Khalad Hasan
SCORE: A Framework for Quantifying Diegesis in Situated Visualization for Augmented Reality
“Augmented Reality (AR) is becoming part of everyday life, yet many AR visuals still appear as floating objects disconnected from nearby surfaces and spaces. Our work helps create more natural, intuitive, and immersive AR experiences by better integrating virtual content into the surrounding environment.”
With the help of Augmented Reality (AR), digital information can be placed directly into the physical world. However, many AR visualizations still feel disconnected from their surroundings, appearing as floating panels or abstract shapes that do not naturally fit into the environment. As AR becomes more common in everyday life, this could lead to visually cluttered experiences that feel distracting rather than immersive.
To address this, we introduce aSCORE, a framework for evaluating how naturally AR visualizations blend into real-world environments. SCORE examines factors such as placement, realism, and environmental consistency to determine whether virtual elements feel like they truly belong. We identified and evaluated key design considerations that can help researchers and designers create AR experiences that are more intuitive, immersive, and seamlessly integrated into everyday life. Overall, our work helps move AR beyond floating digital overlays toward experiences that feel more naturally connected to the real world.
Published April 2026
Satabdi Das
PhD Student, Computer Science
Supervisor: Dr. Mohammad Khalad Hasan
Are You Comfortable Sharing It?: Leveraging Image Obfuscation Techniques to Enhance Sharing Privacy for Blind and Visually Impaired Users
“Privacy is not only about hiding information; it is about giving users control over what they share. For Blind and Visually Impaired users, that control is often limited. Our work takes a step toward designing privacy-aware assistive systems that support confidence.”
Blind and Visually Impaired (BVI) users often share images to seek assistance, connect with others, or preserve memories. However, these images may unintentionally include sensitive content such as identity information, private spaces, medical details, or personal moments.
This research project explores how image obfuscation techniques such as blurring, masking, pixelation, and content based filling can help protect privacy while maintaining comfort in sharing images with family, friends, and strangers.
Our findings show that filtered images significantly improve comfort levels, but no single technique works for every situation. Privacy preferences depend strongly on both the type of content and the audience. This work highlights the need for flexible, user driven, and privacy aware assistive systems for safer image sharing.
Published April 2026
Tanmaya Karmarkar
PhD Student, Computer Science
Supervisor: Dr. Yves Lucet
Computing the convex envelope of bivariate piecewise linear-quadratic functions in linear time
“Engineering and economics solutions involve nonconvex surface optimization —navigating the ‘landscape’ of possibilities full of bumps, ridges, and flat spots for the best answer. Being able to compute the convex envelope rapidly and accurately (rather than guesses) is thus critical and so rewarding. Our work helps provide calculations quick enough for everyday decision-making tools.”
Consider a nonconvex surface in two dimensions composed of quadratic pieces defined over a polygonal partition of the plane. Many real-world optimization problems (in engineering, economics, or machine learning) need the “convex envelope” of such a surface — essentially, the smoothest bowl-shaped surface that sits snugly underneath it. Computing this for two-variable problems has long been difficult. This paper introduces a fast, practical algorithm that builds the envelope step by step, using a clever five-stage process involving mathematical “conjugates.” The authors prove the result has a clean mathematical form and show their method runs in linear time, meaning it scales efficiently as problems grow larger.
Check out their paper in Computational Optimization and Applications.
Published March 2026
Linda Okpanachi
MSc Student, Computer Science
Supervisor: Dr. Ifeoma Adaji
Temporal Analysis of User Engagement, Technology Trends and Emotional Dynamics on Stack Overflow
“This study highlights how developer engagement and emotional dynamics shifted before, during, and after COVID-19. Analyzing different user groups helped reveal long-term patterns that can inform more inclusive and sustainable online technical communities.”
Stack Overflow is a major platform where developers ask questions, share knowledge, and learn from one another. This study examines how user engagement, technology focus, and emotional expression on Stack Overflow changed before, during, and after the COVID-19 pandemic.
By grouping users into regular, intermediate, and expert contributors based on reputation and badges, the study shows that overall engagement declined after COVID-19, with regular users experiencing the sharpest drop. Intermediate users also showed continued disengagement, suggesting a “leaky pipeline” where users struggle to transition from learning to contributing. Expert users remained the most stable contributors and recovered more quickly.
The analysis also reveals shifts toward full-stack development, cloud technologies, and AI, as well as emotional patterns marked by confusion during COVID and increased appreciation and approval afterward. These findings highlight the need for personalized support and long-term strategies to sustain participation in online technical communities.
Published November 2025
Bakdauren-Narbayev
PhD Student, Computer Science
Dr. Mohammad Khalad Hasan
Exploring Pointing and Confirmation Techniques for Teleportation Across Varying Elevations in Virtual Reality
“Forget about holding controllers in Virtual Reality (VR) — explore vast virtual worlds using only your head and hand.”
Moving around in Virtual Reality (VR) can be challenging when the virtual environment is larger than the available physical space. Teleportation is a common solution in which users select a destination and are instantly relocated there. Teleportation typically relies on handheld controllers; however, modern VR headsets can track hand, head and eye movements, enabling more natural interactions in VR. In this work, we evaluated controller-free approaches to teleportation in VR, such as using head and eye movements, hand gestures, and voice commands. Our results showed that the combination of head movements and hand gestures provides the best performance for teleportation. Our findings inform the design of more natural and effective teleportation techniques in VR.
Check out their paper in IEEE Xplore.
Published November 2025
Marium-E- Jannat
PhD Candidate, Computer Science
Supervisor: Dr. Mohammad Khalad Hasan
Exploring Body-Anchored Augmented Reality Interfaces Across Different Mobility and Social Contexts
“Imagine wearing augmented reality glasses and an interface appears on your palm – a body-centered interface that feels more natural to interact with.”
Augmented Reality (AR) devices are becoming more common, letting people see digital information layered on top of the real world through head-worn glasses. Usually, this information appears in front of users, requiring you to reach out into the air to interact with it. In this paper, we explored a more natural way to interact with AR interface – anchor the digital information directly on the body instead. For example, imagine raising your hand and seeing a small menu appear on your palm or forearm to interact with it. We examined how people feel about these body-anchored AR user interfaces in different situations—when they are sitting, walking, or around other people—and how different sizes and layouts affect comfort. We found that people preferred having the AR interfaces on their palm as it felt easy to reach and less tiring. These findings can help shape future AR systems that work more smoothly with our bodies and daily movements, making digital information feel more natural and easier to use in everyday life.
Check out their paper in 2025 IEEE International Symposium on Mixed and Augmented Reality.
Published October 2025
Shawn Zhao
PhD Student, Computer Science
Supervisor: Dr. Fatemeh Fard
Do Current Language Models Support Code Intelligence for R Programming Language? RCR Report
“We built a large-scale R code dataset and showed that current language models struggle with R, highlighting the need and future direction for R developers.”
R is one of the main languages for statistics and data science, but most AI code tools are designed and evaluated mainly on Python, Java, or C. Zixiao (Shawn) Zhao et al. asked whether current language models can really support R developers. The team curated a large dataset from real R packages and evaluated several state-of-the-art models on two tasks: generating natural-language summaries of R functions and predicting method names. They also compared traditional “Base R” code with the tidyverse, a modern R style popular in data science. Overall, the models performed noticeably worse on R than on better-supported languages, especially on tidyverse code and project-specific conventions. The results show that R remains underserved in AI tooling and that dedicated R datasets and models are needed.
Check out their paper in ACM Transactions on Software Engineering and Methodology.
Published October 2025
Manaal Basha
PhD Student, Computer Science
Trust, transparency, and adoption in generative AI for software engineering: Insights from Twitter discourse
“Analyzing 90,000 tweets via topic modeling and open coding, we reveal how users grapple with trust, legal clarity, productivity trade-offs, and safety using AI-based code generation tools, highlighting the need for continuous refinement and clearer guidelines.”
Manaal Basha and Dr. Gema Rodríguez-Pérez analyzed 90,000 tweets using topic modeling, open coding, sentiment analysis, and geographic mapping to explore how developers interact with AI-based code generation tools. Their findings reveal a balance between productivity gains and concerns about trust, legal clarity, and code correctness. Users’ emotions range from enthusiasm to skepticism which vary across Global North and Global South, reflecting cultural and technological contexts in addition to the need for transparency, explainable suggestions, and clear guidance on intellectual property and reliability. This work offers practical insights for tool designers and policymakers navigating human-AI collaboration, and underscores the importance of refining AI tools, providing region-specific guidelines, and establishing legal frameworks for responsible adoption.
Check out their paper in Information & Software Technology.
Manaal also presented this work for an invited talk at the 2025 ACM/IEEE International Symposium on Empirical Software Engineering & Measurement (Hawaii, USA).
Published October 2025
Rishav Banerjee
PhD Student, Computer Science
Supervisor: Dr. Pourang Irani
ThumbSwype: Thumb-to-Finger Gesture Based Text-Entry for Head Mounted Displays
“We introduce ThumbSwype, a one-handed VR/AR text input method where users swipe virtual keys projected on their fingers for fast, ergonomic, smartphone-like typing.”
Text entry on AR and VR devices is notoriously slow and awkward, often relying on midair pointing or clunky virtual keyboards. Rishav Banerjee et al. introduce ThumbSwype, a new method for typing in Virtual and Augmented Reality that mimics the feel and appearance of smartphone swipe keyboards. Users see a typical smartphone keyboard projected onto their index, middle, and ring fingers and type by swiping their thumb across it, just as they would on a phone. In testing, participants reached nearly 15 words per minute, about two-thirds of their phone typing speed, while reporting good ergonomics and usability. ThumbSwype shows how familiar thumb gestures can make typing in immersive environments faster, easier, and far more natural. By bridging familiar mobile interactions with emerging AR and VR systems, this work moves immersive technologies closer to practical, everyday use.
Check out their paper in Proceedings of ACM in Human Computer Interaction.
This paper received an Honorable Mention at the MobileHCI 2025 conference.
Published September 2025
