Algorithms. Decision Support Systems. Deep Learning · Digital Twins
PhD candidate in Computer Science with hands-on experience in computer vision, deep learning, and digital twin systems. Spanning wildfire response, maritime optimization, and biomedical imaging — with a passion for human-in-the-loop AI and XR research.
Working on a NASA-funded project developing a hierarchical AI platform integrated with a digital twin for wildfire mitigation — OpenCV-based visual analytics and human-in-the-loop annotation workflows. Contributing to an ONR-funded project for Navy ship asset allocation using genetic algorithms. Collaborated with NCAR, DRI, CAL FIRE, Clemson, and University of Washington. Designed high-fidelity UE5 extraplanetary digital twin for Moon surface robotics with geospatial terrain integration. Published at WACV 2026, IEEE CAMAD 2025, and AGU 2025.
Developed computer vision-based perception systems for autonomous navigation of TurtleBot3 using ROS2, Nav2, Gazebo, and RViz. Built OpenCV data visualization dashboards for real-time sensor stream analysis.
Taught classes of up to 160 students in data structures, algorithms, and OOP (C++, Java). Conducted research on 5G-enabled telerobotics for economic growth in Bangladesh.
NASA-funded wildfire management platform integrating physics-based fire simulation with interactive 3D analytics. Human-in-the-loop annotation with fire management professionals. Collaborated with NCAR, DRI, CAL FIRE, Clemson.
IEEE CAMAD 2025 ↗High-fidelity lunar terrain with geosynchronization and WGS84 coordinate transforms for robotic mission support and precision navigation to any real lunar location.
OpenCV-integrated visual analytics for real-time risk classification in maritime multi-agent environments with elitist multi-objective optimization (NP-Hard).
Springer 2025 ↗Multi-vessel simulation engine, subconscious navigation learning study, and security awareness framework for naval agents with HCI user studies.
IEEE CoG 2024 ↗CSR-Net crowd density estimation on ShanghaiTech dataset. Real-time crowd violence detection and UAV-perspective crowd tracking pipeline.
IJCNN 2021 ↗CNN-based classification and segmentation for breast cancer detection with comparative analysis of statistical and DL density estimation techniques.
Full pipeline: CAWFE fire prediction model → procedural forest creation → fire spread simulation → interactive analytics → tactical decision making with CAL FIRE professionals.
Vulnerability-aware multi-agent maritime scenario (USS DDG-51) across 243 parameters — an NP-Hard problem. CPA-driven vulnerability mapping enables dynamic adaptive allocation with enhanced collision prevention.
DenseCrowd Turbulence Dataset — simulated UAV crowd videos with Binomial Tracking Algorithm for converging/diverging dense crowd group detection and real-time density estimation.
High-fidelity UE5 lunar terrain with Nanite geometry, 3D tiling, and WGS84 coordinate transforms — enabling real-world positioning and precision navigation to any lunar surface location.
Open to postdoctoral positions, research collaborations, and industry roles in AI, computer vision, and digital twin systems.
✓ STEM-OPT Eligible · I-140 Approved · No Sponsorship Required