PhD student in Computational Science at UC Irvine and San Diego State University. I work on machine learning for materials. My research is on machine-learned interatomic potentials trained on DFT data. Before that I worked on electron microscopy: finding hidden order in 4D-STEM diffraction data and classifying microstructures from SEM images.
Advised by Prof. Mehmet Dogan (SDSU) and Prof. Ruqian Wu (UCI). Based in San Diego.
Resume · CV · Google Scholar · LinkedIn · hashfaq1@uci.edu
San Diego State University · PhD research, Dogan group
- Machine-learned interatomic potentials and DFT calculations. A first project, fine-tuning MACE-MP-0 for monolayer MoS₂, is here.
The Ohio State University, Center for Electron Microscopy and Analysis (CEMAS) · Visiting Fellow, Hwang Lab
- 4D-STEM multislice simulations in collaboration with Sandia National Laboratories.
- Self-supervised and unsupervised learning (K-means, t-SNE) on 4D-STEM nanodiffraction patterns of amorphous semiconductors (Intel-funded SURE fellowship). Poster at Microscopy & Microanalysis 2024.
- Co-author on a manuscript under review at Acta Materialia.
Purdue University · Summer Undergraduate Research Fellow, Chawla group
- ResNet-50 and U-Net++ models that classify Sn-58Bi solder microstructures from SEM images with 90.1% accuracy.
Denison University · Senior capstone
- Automated fold symmetry detection across a 256 × 256 scan of CuZr metallic glass nanodiffraction patterns. Best Senior Capstone Project, Department of Data Analytics.
| Project | What it does | Stack |
|---|---|---|
| Fine-Tuning MACE-MP-0 for MoS₂ | Fine-tunes a foundation interatomic potential on my own DFT data for monolayer MoS₂ and tests it against held-out MD, strain, sulfur vacancies and a from-scratch baseline | Python, MACE, ASE, GPAW, PyTorch |
| Finding Order in Disorder | Detects local 2- to 6-fold symmetry in 4D-STEM patterns of metallic glass, with synthetic validation and a characterization of where rule-based methods break | Python, scikit-image, NumPy, pytest |
| PJM Load Forecast Accuracy | Client project for AEP Energy measuring how PJM demand forecast error changes with horizon, time of week and weather | R, tidyverse, Python, pandas |
| FPN Delivery | Team-built mobile app that connects homebound food pantry clients with volunteer drivers, with live delivery tracking, SMS sign in and an admin dashboard | React Native, Expo, Supabase |
- ML: PyTorch, TensorFlow, CUDA, multi-GPU training, scikit-learn, CNNs (ResNet, U-Net++), self-supervised learning (SimSiam)
- Atomistic modeling: DFT (GPAW, PBE), machine-learned interatomic potentials (MACE), ASE, molecular dynamics
- Scientific computing: NumPy, SciPy, scikit-image, 4D-STEM simulation, HPC at the Ohio Supercomputer Center
- Data and apps: pandas, R and the tidyverse, React Native, Supabase
- PhD, Computational Science, SDSU and UC Irvine Joint Doctoral Program, 2026 to present
- Computer Science and Data Analytics, Denison University, 2026. Denison Phi Society (top 7% of class)