Hi, I'm Isam Balghari
Computational Physicist & Software Engineer
I am a computational physicist and software engineer bridging a six-year industry career in deep learning with an MS in Physics. Working at the intersection of geometric machine learning and condensed matter physics, I develop DSpinGNN—a custom E(3)-equivariant Graph Neural Network that embeds analytical quantum-chemistry biases to scale spin-lattice simulations from 8 to 3,200 atoms, uncovering emergent macroscopic phenomena inaccessible to direct first-principles calculations. To power these physics-informed workflows, I engineered SpinDFT, an automated, high-throughput HPC orchestrator integrating Quantum ESPRESSO, Wannier90, and TB2J, alongside deploying PhysTrackX, an open-source, multi-sensor video-tracking suite for experimental laboratories.
Featured Projects
View all projects →DSpinGNN
An E(3)-equivariant Graph Neural Network combined with a Residual MLP built to predict exchange coupling parameters (J) in magnetic 2D materials. Successfully scaled DFT computations from 8 atoms to large-scale simulations containing 3,200 atoms.
Phystrackx
An independent, proprietary scientific tracking suite designed and deployed for experimental laboratories. Streamlines video tracking, data acquisition, and real-time visualization of complex physical phenomena.
Technical Strengths
Physics & Materials
- • Density Functional Theory (DFT)
- • Quantum ESPRESSO
- • Condensed Matter Physics
- • Spin-Lattice Dynamics
Machine Learning
- • Equivariant GNNs (e3nn)
- • Deep Learning (PyTorch)
- • Geometric Deep Learning
- • Weights & Biases (W&B)
Software Engineering
- • High-Performance Computing (HPC/SLURM)
- • Docker Containers
- • System Architecture
- • Python / TypeScript / C++