PhD Candidate in Electrical & Computer Engineering @ NMSU | AI/ML Research Specialist | Course Creator

Specializing in Power Management & Thermal Optimization, Large Language Models (LLMs), AI Agents, Natural Language Processing (NLP), Machine Learning, and Hardware Security. 7+ years of experience spanning electronic systems design, low-power PCB development, firmware development, and applied AI/ML research. PhD research focuses on leveraging NLP, LLMs, and ML techniques for fine-grained thermal-to-power estimation and dynamic thermal management (DTM) in multicore systems.

• Fine-Grained Power Estimation: Physics-informed neural networks (PINNs) for accurate power profiling in MPSoCs
• Dynamic Thermal Management (DTM): Real-time thermal-to-power estimation for multicore systems
• Blind Power Identification (BPI): DBSCAN clustering + NMF for secure power estimation
• Low-Power Design: Automotive-grade telematics devices, 4-layer PCB architecture optimized for minimal power consumption
• Thermal-Aware Optimization: NSGA-II multi-objective tuning for sub-millisecond inference
• Hardware Validation: NVIDIA Jetson Xavier AGX, heterogeneous SoC testing and benchmarking
• Power/Thermal Sensors: Development for MPSoC attack and defense algorithms (cores + memory subsystems)

• Large Language Models (LLMs): Fine-tuning, prompt engineering, context optimization
• AI Agents: Multi-agent systems, autonomous decision-making, tool integration
• Natural Language Processing: Text analysis, sentiment analysis, language understanding
• Retrieval-Augmented Generation (RAG): Document processing, vector databases, knowledge systems
• Computer Vision: Object detection (YOLO), real-time image processing, driver monitoring
• Anomaly Detection: Statistical methods, adversarial tuning, pattern recognition

• Thermal Trojans: Detection and defense mechanisms for multicore SoCs (77.56% error reduction)
• Power Modeling: Physics-informed models, secure power estimation (84.7% CPU, 73.9% GPU MAE reduction)
• Hardware Security: SoC security, embedded systems protection, malicious sensor attack mitigation
• TinyML: Edge AI deployment, resource-constrained ML systems
• 3D-Stacked Memory Security: Thermal vulnerability analysis in High-Bandwidth Memory (HBM) architectures

• PCB Design: High-frequency circuits, power electronics, signal integrity
• Embedded Systems: ESP32/STM32, FPGA development, IoT systems
• Firmware Development: Real-time systems, communication protocols
• Telematics: Vehicle tracking, industrial IoT, data logging systems

• IEEE HPEC 2024 - High Performance Extreme Computing Conference
  • Cluster-BPI: Efficient Fine-Grain Blind Power Identification for Defending against Hardware Thermal Trojans
• IEEE IGSC - International Green and Sustainable Computing Conference
• IEEE IPCCC - International Performance Computing and Communications Conference
• Concurrency and Computation: Practice and Experience (Wiley)
  • Efficient Deep Learning Models for Brain Tumor Detection
  • Real-Time Control Design and Implementation of Ball Balancer System
• arXiv Preprints - Cutting-edge research in power management, thermal security, and AI/ML
  • Fine-Grained Clustering-Based Power Identification for Multicores
  • Thermal Vulnerability of 3D-Stacked High-Bandwidth Memory Architectures

• CPINN-ABPI: Physics-informed neural network framework achieving 84.7% MAE reduction for power estimation with sub-millisecond inference
• Cluster-BPI: DBSCAN-enhanced blind power identification reducing error rates by 77.56% for thermal Trojan defense
• Thermal Security: Novel detection mechanisms for hardware thermal attacks in multicore SoCs and 3D-stacked memory
• Low-Power Hardware: Automotive-grade telematics devices with FCC/UL/ATEX compliance and minimal power consumption
• Real-time AI Systems: ML-based control systems achieving 99.95%+ accuracy in embedded applications

• Power & Thermal Management: Fine-grained power estimation, dynamic thermal management (DTM), thermal-aware optimization
• Hardware Security: Thermal Trojan detection, malicious sensor attack mitigation, secure SoC architectures
• Physics-Informed ML: Application of PINNs to power/thermal modeling with multi-objective optimization
• Edge AI & TinyML: Resource-constrained ML deployment, real-time inference on embedded systems
• Low-Power Design: Automotive-grade PCB design, power-efficient firmware development

First hardware validation of ABPI on NVIDIA Jetson Xavier AGX, introducing CPINN-ABPI—a hybrid model fusing physics-informed neural networks with the ABPI thermal model and NSGA-II multi-objective tuning.

• Achievement: 84.7% MAE reduction (CPU), 73.9% (GPU), WMAPE 12%
• Performance: Sub-millisecond inference with 85-99% error reduction across simulated heterogeneous SoCs
• Technologies: Physics-Informed Neural Networks (PINNs), NSGA-II optimization, NVIDIA Jetson Xavier AGX
• Application: Real-time power profiling and thermal-aware optimization in multicore systems

Enhanced thermal security framework for multicore SoCs using DBSCAN clustering for improved NMF initialization, refining power estimation and strengthening defense against malicious thermal sensor attacks.

• Achievement: 77.56% error reduction in power estimation
• Innovation: DBSCAN-enhanced BIC framework for hardware Trojan defense
• Technologies: DBSCAN clustering, Non-negative Matrix Factorization (NMF), Bayesian Information Criterion (BIC)
• Application: Securing SoCs from hardware thermal Trojan threats

Enhanced Blind Power Identification (BPI) for multicore SoCs using DBSCAN clustering to improve NMF initialization with steady-state temperature analysis.

• Achievement: 76% error reduction over traditional methods
• Innovation: Improved scalability and precision in thermal management
• Technologies: DBSCAN clustering, NMF, steady-state thermal analysis
• Application: Optimized thermal management in multicore systems

Analysis of convergent thermal-wave attacks in High-Bandwidth Memory (HBM) architectures with compact RC lattice modeling.

• Contribution: Exposed thermal vulnerabilities in 3D-stacked memory
• Technologies: RC lattice thermal modeling, simulation-based validation
• Application: Memory subsystem security in heterogeneous architectures

Compact, automotive-grade telematics and tracking unit designed for low power consumption, real-time sensor acquisition, and vehicle location tracking. Programmed using MicroPython for efficient edge processing.

• Power Design: 9V-30V operation with minimal idle draw
• Features: GNSS/GPS tracking, ignition/battery sensors, flexible I/O (digital up to 30V, analog up to 30V, output up to 50V)
• Connectivity: I²C, Serial, Wi-Fi, Bluetooth, One-Wire protocols
• Firmware: MicroPython for rapid development, low-power control, and OTA updates
• Reliability: FCC-/UL-/ATEX-compliant, automotive-grade rugged design
• Client: Abu Dhabi Police smart city solutions

Classified brain tumors using MR images, comparing InceptionResNetV2, InceptionV3, transfer learning models, and custom BRAIN-TUMOR-net architecture.

• Achievement: Highest accuracy with custom model trained from scratch
• Technologies: Deep neural networks, data augmentation, medical image analysis
• Application: Medical diagnosis and tumor classification

Hybrid pseudo-PD/machine learning algorithm for stabilizing and tracking ball-on-plate system (BOPS) using machine vision.

• Achievement: Servo angle prediction accuracies of 99.95%, 99.908%, and 99.998%
• Technologies: Support Vector Regression, Decision Tree Regression, Random Forest, Fuzzy Logic
• Application: Real-time control systems with ML-based parameter tuning

Comprehensive course on building AI agents with n8n workflow automation.

• Content: 40+ video tutorials, 100+ AI agent implementations
• Features: RAG systems, voice-enabled agents, collaborative agent teams
• Integration: WhatsApp, Telegram, Gmail, and cloud services

Modern ML course with AI-powered approach covering fundamentals to advanced deep learning.

• Content: 19 chapters with PyTorch and Scikit-Learn implementations
• Features: AI-assisted development, Docker containerization, GPU optimization
• Application: Comprehensive ML education with hands-on projects

Power Modeling         │ Physics-Informed Neural Networks (PINNs), Blind Power Identification (BPI)
Thermal Analysis       │ RC lattice modeling, steady-state thermal analysis, dynamic thermal management
Optimization           │ NSGA-II multi-objective optimization, DBSCAN clustering, NMF
Hardware Validation    │ NVIDIA Jetson Xavier AGX, heterogeneous SoC benchmarking
Simulation Tools       │ MATLAB/Simulink, thermal/power profiling frameworks

Large Language Models  │ OpenAI GPT, Claude, Gemini, DeepSeek, xAI, LLaMA
Deep Learning          │ PyTorch, TensorFlow, Keras, Scikit-learn, NumPy, Pandas
Computer Vision        │ YOLO, OpenCV, TensorFlow Object Detection
Specialized ML         │ NVIDIA cuML, Physics-Informed Neural Networks, TinyML
AI Tools               │ n8n, LangChain, Cursor Agents, AutoGen, CrewAI
Vector Databases       │ Pinecone, Weaviate, Chroma, FAISS

Languages              │ Python, MATLAB, C/C++, MicroPython, Verilog/VHDL
Development Tools      │ Cursor, VS Code, Anaconda, Git, Docker, Linux, Jupyter
Cloud Platforms        │ AWS, Google Cloud, Azure, Hostinger
Databases              │ PostgreSQL, MongoDB, SQLite, Redis

PCB Design             │ Altium Designer, KiCad (multi-layer, high-speed, low-power)
Power Electronics      │ LT Spice, P-Spice, power inverter design
Microcontrollers       │ ESP32-S3, STM32, Arduino, PIC, AVR, ARM Cortex M0
Development Boards     │ NVIDIA Jetson Xavier AGX, Raspberry Pi 4, Xilinx Zybo Z7
FPGA Development       │ Xilinx, Verilog/VHDL
Communication          │ UART, SPI, I²C, One-Wire, LIN, CAN
Firmware               │ MicroPython, real-time systems, OTA updates
Compliance             │ FCC/UL/ATEX-compliant design for automotive/industrial applications

Data Processing        │ Pandas, NumPy, SciPy
Visualization          │ Matplotlib, Seaborn, Plotly
Statistical Analysis   │ Regression models (SVR, Decision Tree, Random Forest)
Signal Processing      │ MATLAB/Simulink for hardware modeling and analysis

• PhD Candidate (2023-Present) - Electrical & Computer Engineering, New Mexico State University (NMSU) | GPA: 4.0

  • Research Focus: Power management, thermal optimization, NLP/LLMs for thermal-to-power estimation, hardware security, TinyML
  • Research Assistant: Cutting-edge AI/ML applications in hardware security, SoC power management, computational performance

• Master of Science (2020-2022) - Control and Industrial Power Electronics Application using Machine Learning, Menoufia University | GPA: 4.0

• Bachelor of Electrical & Electronic Engineering (2011-2016) - Faculty of Electronic Engineering, Menoufia University | GPA: 3.81

• AI Plus ME YouTube Channel - AI education and tutorials
  • 40+ comprehensive video tutorials on AI agents and automation
  • Practical implementations of LLMs, RAG systems, and ML algorithms
  • Real-world project tutorials with hands-on coding
  • 1000+ students across courses and tutorials

• Research Assistant (2023-Present) - New Mexico State University, Klipsch School of ECE
• Embedded Hardware Engineer (2022-2023) - Tatweer Company, UAE Government Sector
  • Telematics devices with 4-layer PCB, low-power design for Abu Dhabi Police
  • Smart city solutions: speed radar systems, tracking devices, smart driving test cars
• Research Assistant (2018-2022) - Menoufia University, Industrial Electronics Dept
• 7+ Years Experience - Electronic systems design, low-power PCB development, firmware, IoT, and applied ML

• Power & Thermal Management Research: CPINN-ABPI, Cluster-BPI, fine-grained power estimation frameworks
• AI Agent Projects: Comprehensive n8n automation workflows with 100+ implementations
• Machine Learning Course: Modern ML education with 19 chapters covering PyTorch, TensorFlow, and scikit-learn
• Research Publications: IEEE HPEC, Wiley journals, arXiv preprints on power management and thermal security
• Hardware Projects: Low-power telematics devices, automotive-grade IoT systems, embedded solutions

• Power & Thermal Management: Fine-grained power estimation, dynamic thermal management (DTM), physics-informed neural networks
• Hardware Security: Thermal Trojan detection, malicious sensor attack defense, secure MPSoC architectures
• Edge AI & TinyML: Resource-constrained ML deployment, real-time inference on automotive-grade embedded systems
• Low-Power Hardware Design: Automotive-compliant PCB design, power-efficient firmware, FCC/UL/ATEX certification
• Multi-Objective Optimization: NSGA-II-based tuning for thermal-aware power management with sub-millisecond inference

• 🎥 YouTube: AI Plus ME - AI tutorials and courses
• 💼 LinkedIn: Mohamed Elshamy - Professional networking
• 📚 Google Scholar: Research Publications
• 🔬 ResearchGate: Academic Profile
• 🐙 GitHub: MohElshamy1994 - Open source projects

• Research Partnerships: AI/ML, hardware security, embedded systems
• Industry Consulting: AI implementation, system optimization, technical strategy
• Educational Content: Course development, technical writing, video tutorials
• Open Source Contributions: AI tools, automation frameworks, educational resources

• 1000+ Students across YouTube courses and tutorials
• IEEE HPEC 2024 Publication - Cluster-BPI framework for thermal security
• Multiple Wiley Journal Publications in top-tier venues
• Real-world Deployments of low-power telematics systems for Abu Dhabi Police
• Hardware Validation on NVIDIA Jetson Xavier AGX platform
• Open Source Contributions with active community engagement

• CPINN-ABPI Framework: Physics-informed neural networks for MPSoC power estimation (84.7% MAE reduction)
• Thermal Trojan Defense: Hardware security mechanisms for multicore SoCs (77.56% error reduction)
• 3D-Stacked Memory Security: Thermal vulnerability analysis in HBM architectures
• Low-Power Edge AI: TinyML deployment on automotive-grade embedded systems
• Multi-Objective Optimization: NSGA-II-based tuning for thermal-aware power management

• Advanced Power Management: Next-generation physics-informed models for heterogeneous MPSoCs and chiplet architectures
• AI-Driven Thermal Security: Real-time detection and mitigation of sophisticated thermal attacks in 3D-stacked architectures
• Edge AI for Power Optimization: TinyML deployment for ultra-low-power, thermal-aware embedded systems
• Neuromorphic Computing: Brain-inspired hardware for energy-efficient AI acceleration
• Quantum Machine Learning: Quantum algorithms for power/thermal optimization

• Automotive Systems: Low-power telematics, thermal-aware autonomous driving platforms, smart city infrastructure
• Secure Computing: Hardware-level defense against thermal Trojans, side-channel attacks, and memory-based exploits
• High-Performance Computing: Thermal-aware resource management for data centers and supercomputing clusters
• Industrial IoT: Predictive thermal maintenance, power-efficient edge processing, FCC/UL/ATEX-compliant designs
• Mobile & Embedded: Battery-optimized SoCs, thermal throttling mitigation, real-time power profiling

⭐ Interested in power management research, hardware security, AI/ML collaboration, or consulting? Let's connect and build the future of secure, energy-efficient computing systems!

"Bridging the gap between physics-informed AI and practical power/thermal optimization in real-world hardware"