Computational physicist with six years of industry experience in computer vision and deep learning, now working at the intersection of equivariant machine learning and quantum materials simulation. I build ML architectures from first principles — implementing tensor products, spherical harmonics, and equivariant message passing from scratch — and develop open-source pipelines for materials discovery.
Currently completing my MS Physics at LUMS under Dr. Muhammad Sabieh Anwar and Dr. Muhammad Faryad.
DSpinGNN: Physics-Informed Equivariant GNN for Dynamic Magnetic Exchange in Strain-Deformed CrI₃ Submitted to Physical Review Materials (2026)
Built a bifurcated E(3)-equivariant architecture that simultaneously predicts interatomic forces and instantaneous magnetic exchange couplings J(r), embedding the Goodenough-Kanamori superexchange relationship as an analytical inductive bias. Trained on 8-atom DFT+U cells, deployed without retraining on a 3,200-atom supercell (400× scale transfer, R² = 0.91), extracting mesoscale observables — domain wall width ξ = 1.7 ± 0.3 nm and oscillation period τ = 0.27 ps — inaccessible to direct DFT.
PhysTrackX: Open-Source Framework for Automated Tracking and Sensor Synchronization Apparatus Note in preparation, American Journal of Physics (2026)
An open-source video-tracking and sensor-synchronization framework for experimental physics labs, streamlining data acquisition and analysis pipelines.
| AI for Science | Computational Physics & HPC | Software & Systems |
     
|
     
|
    
|
Audiobooks · Bunjee Jumping · Strategic philosophy · Long, calm drives