Mahmoud Sajjadi

I am an AI researcher and machine learning engineer (Ph.D., Computer Science and Engineering, University of Nevada, Reno) working on foundation models, distributed and federated learning, and mathematical optimization. My core focus is healthcare AI: privacy-preserving, wearable-based methods for detecting early-stage cognitive decline, and LLMs for clinical decision support. I also bring this same machine learning toolkit to power systems, applying federated learning to grid event detection, EV ecosystems, and infrastructure monitoring.

Before moving into AI, I spent nearly a decade as a control and power systems engineer, including data center design and operations. That background lets me build models that work on real sensors, real constraints, and real operations.

Selected results

• 80% less training time for federated and split learning on heterogeneous clients (IEEE Internet of Things Journal, 2024).
• 35% lower communication and storage than LoRA for fine-tuning ViT, LLaMA, and GPT-2 with subspace reparameterization (SOLAR, under review).
• 26% less slower-side compute in decentralized learning with convergence analysis (IEEE ICDCS, 2024).
• 88–90% accuracy on 3,877 power grid events across 23 PMUs with a privacy-preserving federated framework using homomorphic encryption.
• 22% lower energy consumption in infrastructure projects drawing on my prior data center engineering experience.

Areas of expertise

Foundation Models & Machine Learning. Federated and split learning, parameter-efficient fine-tuning (LoRA and subspace reparameterization) for LLaMA, GPT-2, and ViT, and mathematical optimization for distributed training.

Healthcare AI. Multimodal wearable data (PPG, EDA, EEG, IMU) to extract physiological biomarkers and distinguish Mild Cognitive Impairment from healthy cohorts; cloud pipelines on AWS SageMaker; clinical decision-support with LLMs.

Power Systems. Privacy-preserving federated fault detection and PMU event classification across the grid, and EV smart charging and V2G.

I am open to research scientist, ML engineer, and applied AI roles. The best way to reach me is by email.