Experience
Research on the flexoelectric effect in binary metal oxides under the guidance of Prof. Turan Birol.
- Fine-tuning foundational Machine Learning Interatomic Potentials (MLIPs) to enable accurate and efficient modeling of flexoelectricity.
- Discovered that epitaxial strain significantly enhances the flexoelectric coefficient in alkaline-earth-metal oxides by lowering optical phonon frequencies.
- Quantified the relationship between strain and flexoelectric tensor components using first-principles calculations (DFT, DFPT).
- Building a high-efficiency simulation framework enabling accurate modeling of complex crystal structures inaccessible to traditional DFT.
Research on the non-linear Hall effect in Ferroelectrics under the guidance of Prof. Awadhesh Narayan.
- Identified optimal doping concentrations in ferroelectric perovskites that maximize the Berry Curvature Dipole (BCD).
- Systematically varied doping concentrations and computed BCD using Quantum Espresso and Wannier90, demonstrating tunable electronic response.
Teaching Assistant
MatS 4301: Materials Processing Spring 2024- Led laboratory sessions for 25 final-year undergraduate students.
- Demonstrated key materials processing techniques including X-Ray Diffraction, Sputtering, and Dip Coating.
Projects
Financial Fraud Detection Pipeline
Built a robust fraud detection system handling 1 million+ transactions with extreme class imbalance.
- Engineered a hybrid sampling pipeline (Random Under-Sampling + SMOTE) to address the 0.17% minority class problem.
- Tuned an SVM with RBF kernel to achieve 0.93 AUC, reducing false positives by 15% compared to baseline.
- Deployed a K-Means clustering solution that identified 3 distinct fraud patterns previously undetected.
Piezoelectric Modulus Prediction (JARVIS)
Developed an ML pipeline to predict piezoelectric properties for 1,354 inorganic materials.
- Achieved 78% accuracy with an XGBoost classifier to identify piezoelectric materials.
- Utilized SHAP analysis to reveal that crystal symmetry space groups were the dominant predictor.
- Achieved a final RMSE of 0.49 C/m² using a LightGBM regressor integrating a pre-trained band gap model.
Band Gap Classification Framework
Comparative study of ML architectures for separating conductors from insulators.
- Benchmarked Random Forest, SVM, and Neural Networks on a dataset of 5,000 solids.
- Built a classification pipeline identifying XGBoost as the most accurate model.
- Demonstrated that Artificial Neural Networks outperformed all other tested models for continuous band gap regression.
Technical Skills
Scientific Computing & Simulation
Machine Learning
Development & Tools
Education
University of Minnesota
Twin-Cities, US
Doctor of Philosophy - Materials Science
Indian Institute of Science
Bengaluru, India
Master of Science - Chemistry
Indian Institute of Science
Bengaluru, India
BS Research - Chemistry (Major)
Publications & Honors
Selected Publications
- Dominic Varghese, et al.: Flexoelectric effect in alkaline earth metal oxides (Manuscript under preparation)
- Dominic Varghese, et al.: Berry Curvature Dipole in Ferroelectrics (Manuscript under preparation)
- Jana, B., Varghese, D., et al.: Growth of Co9S8 Islands on Cu2S Nano-Disks. J. Phys. Chem. C 2023, 127, 18, 8873–8879.
- Verma, P.K., ..., Varghese, D., et al.: Cobalt(I)-Catalyzed Borylation of Unactivated Alkyl Bromides. Org. Lett., 2020, 22, 1431-1436.
Honors
- NSF ACCESS Discover Grant (2025 - 2026): Resource allocation grant for accessing advanced high-performance computing systems.
- Oral Presentation at APS Global Summit 2025: Presented doctoral research on the Flexoelectric Effect.
- KVPY Fellowship (2018 - 2023): Awarded by the Government of India for exceptional promise in scientific research.
Get In Touch
If anyone wants to work or learn something interesting, feel free to ping me. Open to good interactions and collaborations!
Say Hello