Built an end-to-end mobile AR system that detects objects concealed behind barriers using 60 GHz mmWave radar and visualizes the classified object as an anchored 3D hologram on iPhone. Integrated radar point-cloud acquisition, MATLAB classification, TCP/Wi-Fi streaming, Unity AR rendering, ARKit pose tracking, Blender 3D models, and world-space UI. Designed user-facing AR interactions including scan feedback, object placement, confidence/distance overlays, and tap-to-view information panels.

● Built iPhone AR app anchoring a volumetric photo album onto real-world surfaces using ARKit plane detection. ● Implemented three photo input modes: physical print scanning, camera capture, and iPhone library import. ● Developed crop algorithm for automatic physical photograph detection and digitization. ● Added AR controls for page turning, album customization, and cloud backup via Google Drive. ● Delivered fully functional iOS AR experience with persistent storage and intuitive spatial interaction.

● Developed markerless 3D motion capture pipeline from smartphone video for infant cerebral palsy screening. ● Built full pipeline: smartphone video → DeepLabCut 2D pose → 3D reconstruction → clinical neck-trunk angle extraction. ● Benchmarked three MiDaS depth architectures; selected DPT_Hybrid for optimal accuracy-speed tradeoff. ● Scaled biomechanical body models of infants (22 body parts, joint constraints) in MATLAB Simulink. ● Co-authored research poster published on ResearchGate; presented at NYIT 2024.