Research & Projects
The Data2AI4LifeSciences Laboratory
Dedicated to AI for Life Sciences, grounded in foundational theoretical computer science and applied data science. Research pillars: Data Quality in AI, AI in Life Sciences, and Human Factors.
AI for Life Sciences, Proteomics & Biomedical Imaging
Current & RecentProtGram-DirectGCN
Frontiers 2025A novel method for protein-protein interaction prediction using directed graph convolutional neural networks. This work infers global dense residue transition graphs directly from primary amino acid sequences, enabling accurate interaction screening without requiring physical 3D crystallographic coordinates.
Multimodal Medical Segmentation
NSF Award #2523786Collaborative Research: CISE-MSI project ($600K total / $210K Lead PI) developing auto-prompt textual annotations with domain knowledge for medical image segmentation foundation models. Introduces Spatially Localized Channel Attention (SLCA) validated across clinical imaging benchmarks.
VisInt-X: Visualizing Cross-Linked Proteins
Electronic Imaging
A WebGL and D3-based interactive visualization of data coming out of the XLPM algorithm on 2 protein structures. Presented at Electronic Imaging (EI 2016) and Journal of Imaging Science & Technology.
XLPMOL: OpenGL Protein Model Viewer
OpenGL / PDBAn OpenGL visualization of 2 Protein Models simultaneously based on Protein Data Bank (PDB) data files, allowing comparative structural inspection of protein interactions.
PubMed Knowledge Graph & MedGraph
Nature Sci DataConstruction of a comprehensive PubMed Knowledge Graph with fine-tuned BERT entity extraction, author disambiguation, and NIH grant linkage. MedGraph uses graph representation embeddings (TransE, RotatE) for semantic biomedical search.
ModER: Unsupervised Entity Resolution
Graph ERGraph-based unsupervised entity resolution using Composite Modularity Optimization and Locality-Sensitive Hashing (LSH) for resolving ambiguous records in large-scale knowledge graphs.
CAVE Virtual Reality, 3D Scanning & Graphics
Emerging Analytics CenterCAVE Real-Time Object Scanner
VR-Juggler / RealSense
Real-time application inside the CAVE environment allowing scanning small objects on a custom-made turntable using a RealSense camera to capture mesh data (OBJ), rendered instantly inside the CAVE with Leap Motion manipulation.
Info-Viz in Virtual Environments (Unity)
Creating basic graph elements in Unity game engine to visualize complex relational networks in immersive mobile and head-mounted display (HMD) environments.
Isosurface Extraction from DICOM Volumes
Evaluating algorithms to extract and render isosurfaces from volumetric medical imaging datasets (DICOM) for interactive 3D inspection.
RaymarcherVR
Porting raymarching shader techniques inspired by ShaderToy to desktop VR environments for procedural 3D visual exploration.
Quad Buffered Stereo on Nvidia Quadro
An OpenGL renderer to explore hardware-accelerated quad-buffered stereo rendering on Nvidia Quadro 6000 for immersive stereoscopic displays.
Photosphere & Google Earth in the CAVE
Rendering panoramic images to photospheres and adapting Liquid Galaxy Google Earth KML data visualization within the immersive CAVE environment.
VRJuggler CAVE Interactivity & Point Cloud
Applications demonstrating VRJuggler interactivity with objects and large point cloud rendering inside the CAVE facility.
CUDA Particles in OpenGL
A high-performance simulation and visualization of massive particle systems calculated via CUDA and rendered in real time using OpenGL.
Synchronized Multi-View WebGL Rendering
Synchronizing 3D graphics rendering across several HTML document divs for multi-panel interactive scientific visualization.
Sensor Tracking: Leap Motion & Kinect
Integrating Leap Motion gesture control for CAVE manipulation and presentations, alongside OpenCV-based Kinect calibration and point-cloud generation.
Data Science, Governance & Industry Engineering
Systems & AnalyticsIntel RealSense Visual SLAM GPU
Intel R&DImplemented GPU-accelerated visual Simultaneous Localization and Mapping (vSLAM) using Intel's graphics architecture during my software engineering internship at Intel.
LDA Topic Modeling: Mallet vs. Gensim
Comparative evaluation of Latent Dirichlet Allocation (LDA) topic modeling packages on the 20 Newsgroups benchmark dataset.
Data Governance in Healthcare Information Systems
Big Data governance strategies in hospital environments, presented at UALR Research Expo. Analyzes policy, lifecycle management, and privacy requirements.
Information Security Policies for National Labs
Designed comprehensive security and access policies for the National Visualization Laboratory to safeguard scientific information assets.
Genesis Development Environment (GDE)
MSc Project
A graphical environment for algorithm animation and visual programming in the Genesis language, displaying dynamic data structures and generating executable code.
Intersection Collision Avoidance System (ICAS)
Ain Shams Capstone
Vehicle Infrastructure Integration (VII) client-server application combining Java, GPS, and microcontrollers to link road vehicles to infrastructure and automate collision avoidance.
Conference Paper Review System
A web-based peer review workflow system for conference paper management built with PHP, MySQL, and HTML.
Enterprise Database Architecture & Timesheets
Relational data models, SQL Server triggers/views, a Pet Store Data Warehouse in Microsoft BI Studio, and an interactive GA timesheet management system for Arkansas Tech University.