Hello, I'm
Mahmoud Sameh
AI Researcher & Engineer
Electrical engineer with a deep focus on AI. I research, build, and ship. My work spans computer vision, foundation model adaptation, and applied ML across domains from fintech and healthcare to robotics.
Publications
Real-Time Human Fall Detection from Video Using YOLOv11 with Pose Estimation: A Paradigm Shift Towards Efficient Transformer-Based Architectures
A. BenAbdennour, M. Sameh, B. A. Khawaja, A. K. Vallappil, A. M. Alenezi, Q. H. Abbasi, S. Qazi
This paper presents a real-time fall detection system leveraging YOLOv11 with integrated pose estimation. By adopting an efficient transformer-based architecture, the system achieves high detection accuracy and low latency suitable for deployment in clinical and home monitoring environments, representing a significant advance over prior CNN-based approaches.
Real-Time Human Fall Detection from Video Using YOLOv11 with Pose Estimation: A Paradigm Shift Towards Efficient Transformer-Based Architectures
A. BenAbdennour, M. Sameh, B. A. Khawaja, A. K. Vallappil, A. M. Alenezi, Q. H. Abbasi, S. Qazi
This paper presents a real-time fall detection system leveraging YOLOv11 with integrated pose estimation. By adopting an efficient transformer-based architecture, the system achieves high detection accuracy and low latency suitable for deployment in clinical and home monitoring environments, representing a significant advance over prior CNN-based approaches.
Predicting Emergency Department Patient Arrivals at Hospitals Using Machine Learning Techniques
A. M. Alenezi, M. Sameh, M. Aljohani, H. S. Alharbi
Accurately forecasting emergency department (ED) patient arrivals is critical for hospital resource planning and reducing patient wait times. This study applies and compares multiple machine learning techniques to predict daily and hourly ED arrival volumes, enabling proactive staffing and capacity management.
Predicting Emergency Department Patient Arrivals at Hospitals Using Machine Learning Techniques
A. M. Alenezi, M. Sameh, M. Aljohani, H. S. Alharbi
Accurately forecasting emergency department (ED) patient arrivals is critical for hospital resource planning and reducing patient wait times. This study applies and compares multiple machine learning techniques to predict daily and hourly ED arrival volumes, enabling proactive staffing and capacity management.
Efficiently Adapting SAM 2 for Automated Schematic Capture from Hand-Drawn Circuit Diagrams
M. Sameh, A. BenAbdennour, J. K. Ali
This paper proposes an efficient adaptation of Segment Anything Model 2 (SAM 2) for the task of automated schematic capture from hand-drawn electrical circuit diagrams. By fine-tuning SAM 2's video segmentation capabilities on circuit imagery, the system accurately segments and identifies components, enabling downstream netlist extraction and circuit simulation.
Efficiently Adapting SAM 2 for Automated Schematic Capture from Hand-Drawn Circuit Diagrams
M. Sameh, A. BenAbdennour, J. K. Ali
This paper proposes an efficient adaptation of Segment Anything Model 2 (SAM 2) for the task of automated schematic capture from hand-drawn electrical circuit diagrams. By fine-tuning SAM 2's video segmentation capabilities on circuit imagery, the system accurately segments and identifies components, enabling downstream netlist extraction and circuit simulation.
Featured Projects
Simak: Agentic Arabic-First Document Extraction
2026An agentic, Arabic-first document-understanding platform for the Saudi market. Non-technical users describe the document they need to extract, an AI agent interviews them, proposes a JSON schema, and deploys a structured-output LLM extractor behind an API key.
Simak: Agentic Arabic-First Document Extraction
2026Simak (سماك, “pillar/support”) turns plain-language requirements into production document extractors. A user describes what they need, a wizard agent interviews them, proposes a JSON schema, and deploys an extractor powered by structured-output LLMs, all behind a simple API key.
Built Arabic-first for Saudi document workflows, with PDPL compliance and bilingual RTL support as hard constraints.
FragLibrary: An AI-Powered Fragrance Discovery Platform
2025Full-stack fragrance discovery platform (10,000+ fragrances) with a custom-trained embedding model (Scent2Vec 2.0) that maps scent preferences to closest matches via cosine similarity.
FragLibrary: An AI-Powered Fragrance Discovery Platform
2025Built and deployed a full-stack fragrance discovery platform covering 10,000+ fragrances. Trained a custom embedding model (Scent2Vec 2.0) that encodes scent descriptors into a vector space, enabling cosine-similarity-based preference matching. Deployed on Vercel with a Django REST backend and PostgreSQL database.
CircuitVision: AI-Powered Recognition of Hand-Drawn Electrical Circuits
2025End-to-end system converting hand-drawn circuit diagrams into simulatable netlists via fine-tuned YOLOv11 detection, SAM 2 segmentation, custom OpenCV connectivity, and Gemini API for value recognition. Awarded 2nd Best Graduation Project in AI university-wide.
CircuitVision: AI-Powered Recognition of Hand-Drawn Electrical Circuits
2025Led development of an end-to-end pipeline that converts hand-drawn circuit diagrams into simulatable netlists. The system chains fine-tuned YOLOv11 for component detection, custom OpenCV algorithms for connectivity analysis, SAM 2 for segmentation, and the Gemini API for component value recognition. Awarded 2nd Best Graduation Project in AI university-wide at Islamic University of Madinah.
Career Journey
From foundations to AI, a timeline of growth.
AI Researcher & Consultant
Independent
Conducting independent AI research and taking on select consulting projects across applied ML, computer vision, control, and embedded systems.
AI Engineer Intern
Aajil Fintech
Built AI infrastructure to automate credit risk assessment. Developed a feature-engineering pipeline generating financial analyses from raw application data. Trained and deployed classification/extraction models with end-to-end MLOps.
Research Intern
King Fahd University of Petroleum and Minerals
Designed a complete experimental protocol for a novel hyperspectral imaging dataset of date fruits. Developed the full software stack for push-broom HSI hardware and created an open-source GUI (PyQt6) now under journal review at SoftwareX.
B.S. Electrical Engineering
Islamic University of Madinah
Chose EE for its balance of hardware, software, and AI. Graduated top of class (4.95/5). Complemented formal education with Harvard's CS50 and self-driven courses in Python, C, and AI. Led and developed numerous projects across computer vision, robotics, and applied ML. Led technical aspects of multiple research papers as first or second author, won Enjaz Hackathon, earned 2nd Best Graduation Project in AI, and passed KAUST Academy's AI stages.
Preparatory Year
Islamic University of Madinah
Passed with a perfect 5/5 GPA without much effort, thanks to the self-study foundation built during the gap year (Thank you Khan Academy). Used the abundant free time to go deeper into calculus, linear algebra, and coding through online courses.
The Gap Year
Self-Directed
Took a deliberate year off to relearn mathematics the right way, revitalizing a love that high school methods had nearly killed. If you hate math, you're learning it wrong. Dove into JavaScript, computer science fundamentals, and psychology. Khan Academy was the star of that year: 1.9 million energy points earned through relentless dedication. A year where my love for coding and computers grew rapidly.
High School Graduate
Top 5 in Class
Graduated as one of the top 5 students. Left with a determination to learn differently, setting the stage for everything that followed.