• The project aims to computationally study adsorption of asphaltene molecules on calcite surface using MD simulations.• Developed a strong conceptual understanding of molecular simulations by reviewing relevant literature.• Learned to perform molecular dynamic (MD) simulation in GROMACS software on High-performance computing (HPC) cluster.• Lab name: Materials Modeling and Simulation Laboratory

• Reproduced results from Ant Wall model paper by Professor Prateek Jha, which aims to model the release of drug molecules from the polymer matrix using lattice-based computer simulation techniques.• Applied concepts of Monte Carlo sampling, Object Oriented Programming and Python-based post-processing of the simulation results. Random numbers were generated using the GNU Scientific Library for random sampling.• Visualized the change in concentration of drug molecules in the matrix with the number of Monte Carlo steps.• Currently working on extending the model to incorporate more complex phenomena, such as multilayered drug release involving a hydrophobic and hydrophilic layer.• Lab name: Molecular Simulations and Drug Delivery Research Group

• The objective of the project was to reduce the computational cost of simulating spirally wound Lithium-ion cells by coupling it with an equivalent flattened, unrolled cell domain using coordinate mapping based methods. • Conducted an extensive Literature review of existing work on MultiScale-MultiDomain (MSMD) and Multi-Dimensional Modelling (MuDiMod) framework which aligned with our research objective. • Established a strong understanding of the physics based modelling of Lithium ion cell. • Created models on COMSOL Multiphysics software to validate our findings • Lab name: Energy devices and Multiphysics Lab

• Worked on identification and quantification of voids which was found in our material of interest. The given dataset consisted of high resolution Computed Tomography (CT) scan images of the equipment. • Applied the U-net architecture for segmenting out the voids from the rest of the background. The model was trained using DICE loss as the loss function. Achieved a Dice coefficient of 0.96. • Curated a data pipeline for preprocessing data and generating insights using the trained model to analyze the samples. • Created a tool for hand annotation and ground truth mask generation of voids required for training the model. • Github Link - https://github.com/HarshitBuilds/Research-Internship-in-Computer-Vision.git

Assisted first-year students in their course on Quantum Mechanics and Electromagnetic Waves