AI & ML interests
The TCCDEM research group focuses on advancing computational materials science and crystal structure prediction (CSP) by deeply integrating machine learning and deep learning into materials discovery, particularly through the development of frameworks like PNcsp and PNcsp+ that utilize chemical similarity based on Mendeleev's Periodic Number representation. Key projects involve building end2end automated pipelines that interface local database querying with high throughput screening across repositories like OQMD, the Materials Project to systematically scan chemical systems and discover novel stable phases. Furthermore, the group leverages SOTA machine learning interatomic potentials and graph neural network architectures similar to MACE, M3GNet, ALIGNN-FF and CHGNet alongside advanced deep learning architectures such as large language models, enabling rapid structure prediction, relaxation, automated property matching, space group selection and smart data filtering pipelines that bypass the traditional computational bottlenecks of Density Functional Theory.