Projects

GranularNet: Physics-Informed Neural Networks for Granular Segregation

Fall 2025 · Department of Mechanical Engineering, University of Michigan

• Formulated and solved the advection–segregation–diffusion PDE for granular flows using Physics-Informed Neural Networks (PINNs), enabling mesh-free prediction of spatiotemporal concentration fields.
• Implemented forward PINN models in PyTorch to reproduce continuum segregation behavior consistent with DEM simulations and experiments.
• Developed inverse PINN frameworks to learn unknown physical parameters (e.g., segregation length scale Λ) and extended constitutive models directly from sparse data.
• Designed and trained neural network closures to discover data-driven segregation velocity laws, replacing traditional empirical models while enforcing physical constraints.
• Achieved high predictive accuracy (RMSE ≈ 10⁻³) and demonstrated strong agreement with ground truth data across multiple regimes and flow conditions.
• Analyzed model performance across different parameterizations, highlighting trade-offs between interpretability (parametric models) and flexibility (neural closures).

Molecular Dynamics Simulation of Shock-Compressed Nickel

Winter 2025 · Department of Mechanical Engineering, University of Michigan

• Implemented piston-driven shock simulations in LAMMPS using Embedded Atom Method (EAM) potentials to model high strain-rate deformation in FCC nickel.
• Simulated shock propagation across multiple regimes (elastic, elastic-plastic, overdriven) by varying piston velocities and initial temperatures (5–600 K).
• Analyzed shock wave structures, including elastic precursor and plastic wave splitting, and quantified shock speed–particle velocity (Us–Up) relationships.
• Performed microstructure characterization using OVITO with Common Neighbor Analysis (CNA) to identify phase transformations (FCC → HCP/BCC) and stacking fault formation.
• Evaluated stress evolution using per-atom virial stress and von Mises stress, linking atomic-scale mechanisms to continuum-level plasticity and Hugoniot elastic limits.
• Investigated temperature effects on shock response, demonstrating reduction in elastic wave speed and enhanced phase transformation at elevated temperatures.

Parallelized Object-Oriented Framework for MCMC and MD Simulations

Fall 2024 · Department of Mechanical Engineering, University of Michigan

• Designed and implemented an object-oriented Monte Carlo (MCMC) simulation framework for modeling molecular systems.
• Developed parallelized MCMC simulations using OpenMP for multi-threaded execution and MPI for distributed computing across multiple nodes.
• Conducted parallel performance analysis of Molecular Dynamics (MD) simulations using LAMMPS, comparing CPU and GPU efficiency.
• Evaluated parallel performance scaling using strong and weak scaling tests to assess computational efficiency.

Computational Study of Argon Atoms Using Monte Carlo and Molecular Dynamics Simulations

Fall 2024 · Department of Mechanical Engineering, University of Michigan

• Completed canonical ensemble (NVT) simulations using Metropolis-Hastings algorithm in MCMC and Lennard-Jones potential in MD.
• Analyzed radial distribution function (RDF) to quantify structural changes in Argon at different state points.
• Conducted diffusion coefficient analysis using the Einstein relation and validated results through comparison with literature values.
• Explored computational efficiency by scaling MD simulations on multi-processor systems, optimizing performance on high-performance computing clusters.
• Utilized LAMMPS, Python, and C++ for simulation implementation, data analysis, and performance optimization.

Heat Conduction Modeling Using Physics-Informed Neural Networks

Fall 2023 · Department of Mechanical Engineering, University of Michigan

• Implemented self-adaptive weighting techniques to improve model accuracy and enforce boundary conditions effectively.
• Utilized TensorFlow 2 and Adam optimizer to train neural networks, integrating governing physical laws into deep learning models.
• Compared PINN results with finite difference method (FDM) solutions to validate accuracy.
• Achieved 93.2 % accuracy in predicting the thermal diffusivity coefficient of an unknown material through physics discovery.

Machine Learning Application in Solving Partial Differential Equations

Fall 2023 · Department of Mechanical Engineering, University of Michigan

• Developed and implemented neural network models to solve PDEs, focusing on integrating uniquely designed cost functions to enforce boundary conditions and physical laws.
• Analyzed the effectiveness of ML approaches compared to conventional FEM solutions, highlighting potential advantages in computational efficiency and adaptability.
• Conducted experiments to compare the performance of machine learning-based solvers with classical methods like finite difference and finite element methods, showing reduced computational complexity compared to traditional numerical methods.

Master of Applied Science Thesis

2021–2023 · Composites Research Network, The University of British Columbia, under supervision of Prof. Abbas S. Milani

• Experimental Material Characterization and Hemisphere Forming Process
• Development of a Finite Element Framework as a Reference for Validation
• Implementation of Material Point Method for Numerical Characterization of Materials and Modeling of Hemisphere Forming Process

Finite Element Modeling of Forming Processes and Numerical Validation of Material Characterization

Spring 2022 · School of Engineering, UBC

• Development of a Fortran Subroutine for Woven Fabrics Constitutive Model in ABAQUS
• Development of Python Scripts for Numerical Modeling of Hemisphere Forming Process in ABAQUS
• Numerical Validation of In-Plane (Tension and Shear) and Out-Of-Plane (Bending) Behavior of Woven Fabrics

Experimental Material Characterization and Forming Process

Fall 2021 · School of Engineering, UBC

• Hemisphere Forming Process and Micro-CT 3D Imaging
• Tensile, Bias Extension, Picture Frame, Peirce Bending and Inter-Ply Friction Test

Numerical Modeling of Pultrusion Process

Spring 2021 · School of Engineering, UBC

• Numerical Modeling and Validation of Heat Transfer in Pultrusion Process
• Numerical Modeling and Validation of Kinetic Reaction in Pultrusion Process

MITACS Internship at NIEDNER INC.

2022 · NIEDNER INC., Coaticook, Quebec, Canada

• Advanced experimental material characterization of 3D woven fabric composite materials, by conducting tensile, bias extension, peirce bending, inter-ply friction tests, and Micro-CT 3D imaging.
• Numerical analysis of buckling-driven in-situ instabilities using the finite element method and material point method.

AIAA Engine Design Competition

2018–2019 · Department of Aerospace Engineering, Sharif University of Technology

• Engine Cycle and Axial Turbine Design
• Mechanical Stress and Vibration Analysis (Natural Frequencies, Campbell Diagram, etc.)

U.S AAS CanSat Design and Build Competition

2017–2018 · Department of Aerospace Engineering — Stephenville, Texas

• Mechanical Subsystem Design
• Descent Control Design

Bachelor of Science Thesis

Spring 2020 · Department of Aerospace Engineering

• Deriving Governing Equations Using Hamilton's Principle and Ritz Method
• Aeroelastic Instability Analysis Using Piston Method

Design and Analysis of a Pressure Vessel, Saddle and its Platform

Spring 2018 · Department of Mechanical Engineering, SUT

• Pressure Vessel and its Connections and Reinforcements Design Using ASME BPVC VIII
• Pressure Vessel Static Structural Finite Element Analysis Using ANSYS
• Platform and Holder Design Using Shigley's Methodology
• Non-permanent (Fasteners) and Permanent (Welding) Joint Design and Analysis Using Shigley's Methodology

Design and Analysis of a Two-stage Reduction Power Transmission (Gearbox)

Fall 2018 · Department of Mechanical Engineering, SUT

Character Recognition of Handwritten Number Images Using a Machine Learning Algorithm

Summer and Fall 2018 · Internship project, MAPNA Turbine Engineering and Manufacturing Company (TUGA)

• Developing a General MATLAB Code and a GUI for Convolutional Neural Networks
• Testing and Validating with MNIST Database

Character Recognition of Seven-segment Numbers Using a Machine Learning Algorithm

Summer 2018 · Internship project, MAPNA Turbine Engineering and Manufacturing Company (TUGA)

• Developing a General MATLAB Code for Deep Neural Networks
• Testing and Validation

CCHP Cycle Design for the Department Building

Fall 2017 · Department of Aerospace Engineering

• Energy Consumption Estimation for the Department Building
• Cycle Design and Analysis Using Aspen HYSYS

Design and Analysis of a Power Transmission (Gearbox) Components

Fall 2018 · Department of Mechanical Engineering

• Shafts Design and Calculations
• Bearings Selection
• Pulley and Belt Design
• Chain and Sprocket Design
• Components Modelling Using Solidworks

Design and Analysis of a Power Transmission (Gearbox) Shafts

Spring 2018 · Department of Mechanical Engineering

• Shafts Design and Calculations
• Bearings Analysis
• Components Modelling Using Solidworks

Design and Analysis of a Truck Weighbridge

Spring 2018 · Department of Mechanical Engineering

• Mechanical Springs Design and Analysis
• Platform Design and Analysis
• Components Modelling Using Solidworks