Senior computer engineering student at the University of Sharjah (3.91 CGPA). I build lightweight, multi-sensor perception for edge devices and test it on data the model has never seen.
- Aerial search and rescue. RGB and thermal person detection for UAVs, with causal temporal post-processing to reduce false alerts. Accepted at ICSPIS 2026 (Dubai, 10–12 November).
- Driver monitoring. A subject-disjoint benchmark of lightweight detectors on RGB and near-infrared video. Conditionally accepted at CAISAIS 2026; revision in progress.
- MANAR. A supervised-autonomy search-and-rescue drone. The C++ control core and edge person detector are built; temporal multisensor fusion is the next milestone.
| Project | Area | What it is |
|---|---|---|
| multi-modal-detection | Edge AI / UAV perception | RGB and thermal person detection with temporal post-processing for aerial search and rescue. Apache-2.0. |
| lightweight-driver-behavior-detection | Edge AI / computer vision | Subject-disjoint benchmark of lightweight detectors for driver warning cues on RGB and NIR video. Apache-2.0. |
| manar-search-rescue-drone | Robotics / sensor fusion | Search-and-rescue drone combining RGB/IR/thermal vision, FMCW radar, passive RF, and audio. Proprietary license. |
| macroblock-visual-automation | Desktop automation | Scratch-style Windows automation with a mandatory global abort hotkey. MIT. |
| class-watcher-uos | Student utility | Course-seat monitor for University of Sharjah registration with phone, desktop, and email alerts. |
- Languages: Python, C++, TypeScript
- ML and vision: PyTorch, Ultralytics YOLO, D-FINE, RT-DETR, OpenCV, NumPy, Mamba state-space models
- Desktop: PySide6, Tkinter
If you work on edge AI, radar perception, or applied signal processing, connect on LinkedIn.