Scroll to play the story. Chapters shift the 3D world behind the UI.
Multidisciplinary Full-Stack MERN developer and AI engineer specialized in deep learning, computer vision, and NLP. Experienced in building scalable web applications, real-time video intelligence systems, AI-powered platforms, and large-scale data pipelines—from data ingestion and preprocessing to model evaluation, deployment, and UI integration.
Click skill-cards to reveal what I use to build products, pipelines, and real-time systems.
The highest-signal projects — built end-to-end, measured, shipped.
Correcting noisy Arabic text using Transformer-based architecture (T5) with a large-scale scraped dataset pipeline.
End-to-end real-time behavioral video analysis platform supporting RTSP streams and uploaded video. Implemented dual frame selection strategies (Skip Frame and Motion-Adaptive) to optimize inference performance under real-time constraints.
Real-time school bus tracking and management platform with dashboards for drivers and parents.
Augmented Reality system allowing users to preview clothing items in real time using camera input and 3D clothing assets.
Product management system with 14+ backend endpoints and a frontend state store aligned with backend logic.
Experiments comparing Skip Frame Selection and Motion-Adaptive Frame Selection techniques to reduce computational cost while maintaining classification performance.
Everything I’ve built (or helped build), cataloged.
Correcting noisy Arabic text using Transformer-based architecture (T5) with a large-scale scraped dataset pipeline.
End-to-end real-time behavioral video analysis platform supporting RTSP streams and uploaded video. Implemented dual frame selection strategies (Skip Frame and Motion-Adaptive) to optimize inference performance under real-time constraints.
Experiments comparing Skip Frame Selection and Motion-Adaptive Frame Selection techniques to reduce computational cost while maintaining classification performance.
Comparative analysis of frame selection strategies for video classification using multiple 3D CNN architectures.
Real-time human pose classification pipeline combining YOLOv5 for detection and LSTM for sequence-based classification.
Real-time school bus tracking and management platform with dashboards for drivers and parents.
Product management system with 14+ backend endpoints and a frontend state store aligned with backend logic.
AI-powered job hunting automation system prototype using no-code workflows and API integrations.
Augmented Reality system allowing users to preview clothing items in real time using camera input and 3D clothing assets.
Web-based platform for live 3D mesh preview and management using MERN stack architecture.
Interactive cursor tracking system using ReactiVision and TUIO protocol with path rendering and multi-cursor support.
Flutter development environment configuration, emulator performance diagnosis, and Android tooling troubleshooting.
Expanded structured datasets from hundreds to thousands+ records for training and experimentation.
Concept platform for virtual property viewing with AI-driven similarity detection and recommendations.
Proof of work, momentum, and range.
How I design under real-world constraints: latency, scale, security, and maintainability.
Email me or reach out via social links.