Postdoctoral Researcher · NC State University
Developing machine learning interatomic potentials for reactive chemistry, bridging ab initio accuracy and large-scale molecular dynamics.
I'm Rahul, a postdoctoral researcher in theoretical and computational chemistry at NC State University, where I work in the Pfaendtner Research Group. My research applies machine learning to make molecular simulations faster, more reliable, and easier to use for studying complex chemical processes. I lead the development of SPARC (Smart Potential with Atomistic Rare Events and Continuous Learning), an open-source Python toolkit for training interatomic potentials.
I completed my PhD in chemistry at the Prof. Nisanth Nair Lab (IIT Kanpur), where I worked on the applications of QM/MM techniques to model reactions in zeolites. Outside research, I enjoy playing badminton and watching movies.
Open-source Python toolkit automating query-by-committee active learning for reactive machine learning potentials with ASE, PLUMED, and AIMD relabeling.
Review of computational approaches for zeolite catalysis, covering QM/MM, AIMD, and emerging machine learning methods.
Mechanistic and free-energy analysis of proton-exchange pathways in acidic zeolite frameworks.
Combined experimental kinetics and DFT to unravel bimetallic hydrodeoxygenation mechanisms.
Designed a local antibiotic delivery system that prevents biofilm formation.
Correlated quantum methods applied to ammonium ion hydration using a molecular tailoring approach.
Small-aggregate analysis to decode packing motifs in carbon disulfide crystals.
Languages and tools reflected across my GitHub repositories, including SPARC, DeePMD/LAMMPS build workflows, and enhanced-sampling analysis codes.
Open-source active learning workflow for training reactive machine learning interatomic potentials, published in the Journal of Open Source Software.
Modular Fortran code I wrote to reweight Temperature Accelerated Sliced Sampling (TASS) output from CPMD/PLUMED runs. Computes multidimensional free energy surfaces via WHAM reweighting or the mean force method, with B-spline interpolation support.
Python package to center molecular structures within periodic boundary conditions and convert XYZ trajectories
into VASP (POSCAR) or Quantum ESPRESSO input files via the xyzcenter and xyzconverter CLI tools.
Installable Python package (LJengo) for Lennard-Jones molecular dynamics with NVE and NVT ensembles,
including force calculations, trajectory propagation, and energy plotting utilities.
Hands-on notebook for the SPARC active learning workflow: building machine learning interatomic potentials for reactive systems with automated sampling, relabeling, and model refinement.
Workflow: Active learning loop with ML/MD → query-by-committee → ab initio relabeling → retrain
Tutorial for training a NequIP machine learning potential on a 22-water cluster in a periodic box. Covers data generation with Quantum ESPRESSO, model training, and LAMMPS-based molecular dynamics.
Workflow: 00.data → 01.train → 02.lmp
DeepModeling-style tutorial for building a DeePMD potential and exploring the 1,3-butadiene cyclization mechanism. Uses VASP-generated training data, DeePMD-kit training, LAMMPS MD, and PLUMED enhanced sampling.
Workflow: 00.data → 01.train → 02.lmp → 03.plumed
AIR-42 in Joint CSIR-UGC Exam (Chemical Sciences)
All India Rank 42 among national research fellowship candidates.
Junior Research Fellow
Funded doctoral research in computational chemistry.
Senior Research Fellow
Continued fellowship support through PhD candidacy.
CECAM 2026
Cornell Tech, New York
LAMMPS 2025
Albuquerque, New Mexico
FOMMS 2024
Utah, USA
TCS 2021
IISER Kolkata
RARE-21
IIT Kanpur
APATCC-9
University of Sydney
ML for Science
IIIT Hyderabad
CRSI National Symposium in Chemistry
Workshop FECCBS-2019
IIT Kanpur
APCTCC-8
IIT Bombay