Current and Past Research Students
-
Danrong Zhang, Ph.D., CSE, Georgia Tech
-
Jiaojun Liu, Ph.D., CEE, Georgia Tech
-
Karie Yamamoto, Ph.D. Candidate, CEE, Georgia Tech
-
Bharath Raghunathan, MSCS, Georgia Tech
-
Oleksandr Horielko, BSCS, Georgia Tech
-
Sakshi Deshpande, MSCS, Georgia Tech
-
Ryan Doan, BSCS, Georgia Tech
-
Ryan Zhang, BSCS, Georgia Tech
-
Huaijin Tu, BSCS, Georgia Tech
Research Areas
Artificial Intelligence in Computing Education
My research explores how AI can enhance learning across a student’s entire journey through the computing curriculum. For early-stage students in CS1 and CS2, I design and evaluate AI-powered tools that support personalized learning at scale. These include real-time feedback systems that track syntactic, conceptual, and strategic coding errors, as well as Socratic hint generation techniques that promote self-reflection and deeper understanding without short-circuiting the learning process. By leveraging LLMs, these tools help identify struggling students and guide instructional focus on the most relevant topics in large-enrollment courses.
As students progress to more advanced coursework, particularly in Software Engineering and Capstone courses, my work shifts to empowering them to use Generative AI tools critically and effectively. These interventions focus on how students integrate tools like GitHub Copilot, Codeium, Claude, Lucidchart AI, Mabl, and Testim across different phases of the software development lifecycle—including design, implementation, debugging, and testing.
My studies examine student perceptions, productivity, and the challenges of hallucinations, over-reliance, and prompt engineering, helping students develop both technical fluency and critical awareness of AI’s role in software engineering workflows.Through this research, I aim to scaffold AI-enhanced learning in a way that evolves with students’ growing expertise, preparing them to be thoughtful and capable users of AI in both academic and professional settings.
Sustainability in Computing Education
As computing continues to impact global energy consumption, integrating sustainability into the CS curriculum has become essential. My work in this area includes the redesign of multiple software engineering courses to embed sustainable software engineering principles. Students engage with energy-efficient coding, modify UML diagrams to incorporate sustainability trade-offs, and apply ethical decision-making frameworks during project-based work.
Through targeted interventions, students gain hands-on experience with carbon-conscious computing and lifecycle-aware system design. My research also explores student reflections, learning outcomes, and behavioral shifts in response to sustainability-themed assignments. By aligning curriculum with the United Nations Sustainable Development Goals (SDGs), this work empowers students to see software not just as a technical product, but as a tool for broader societal and environmental impact.
AI-Driven Geotechnical Intelligence and Disaster Resilience
In collaboration with civil and environmental engineers, I apply machine learning and computer vision techniques to address pressing challenges in geotechnics and disaster response.
This interdisciplinary research includes:
-
Landslide susceptibility mapping using positive-unlabeled learning to improve predictions in high-risk, data-scarce regions like the Himalayas.
-
Post-disaster damage assessment leveraging AI and surrogate datasets (e.g., satellite imagery, nighttime light intensity) to identify infrastructure damage and power outages in near-real-time.
-
Optimization of geotechnical instrumentation using machine learning to refine the design of multi-sensor cone penetrometer devices, reducing sensor complexity without compromising classification performance.
This work bridges data science, civil engineering, and AI to improve the resilience and safety of infrastructure systems in the face of natural hazards.
Multi-Scale Characterization of Geomaterials and Pore Structures with Computational and Deep Learning Approaches
My research focuses on the multi-scale characterization of geomaterials, with an emphasis on understanding the geometry, connectivity, and behavior of three-dimensional pore structures in particulate materials. These microstructural features are critical for applications ranging from subsurface soil analysis and petroleum extraction to biomedical materials and porous metals.
A significant portion of this work involves developing operator-independent algorithms to segment and quantify the 3D pore space into physically meaningful elements such as pore bodies and throats. These geometric insights enable the study of how material microstructure influences macroscopic responses under static and dynamic loading.
To extend these insights to large-scale systems, I also design physics-informed deep learning workflows capable of rapidly predicting a wide array of microstructural properties from 3D image data. These workflows capture both individual features (e.g., particle size, throat area) and collective metrics (e.g., tortuosity, connectivity, anisotropy), offering scalable characterization of geomaterials across resolutions.Together, these computational and AI-driven methods provide a foundation for understanding and engineering particulate systems in energy, civil infrastructure, and materials science domains.
Micro-to-Macro Analysis of Shear-Induced Deformation of Granular Materials
The macroscopic behavior of granular materials stem from a heterogeneous interplay of microstructural responses and is influenced by micro-mechanical descriptions of particles, their interconnected network of contacts, and associated pore bodies and throats.
This work explores the role of micro-mechanical properties in providing insights into the deformation characteristics of sheared granular sands from the perspective of pore space, which has been overlooked for a long time as compared to particle and contact measurements.
Analyzing the complex micromechanics of bio-cemented soils
Bio-cementation improves the mechanical behavior of the subsurface by integrating bacterial metabolic activity in soil structure to induce the formation of calcite. The spatial location and distribution of the calcite bonds have a huge effect on the macroscopic improvement in the behavior of the cemented material.
​
This work explores the role of 3D micro-mechanical characterizations of cemented soil microstructures to gain insights into the factors affecting improvement in macro response to learning ways to engineer the treatment methods for better results.