PhD in Systems Science — Computational Mechanics

Thesis: Applications of Physics-Informed Neural Networks (PINNs) in CFD

Advisors: Andrea Mola, Giovanni Stabile

Courses: Computational Fluid Dynamics for Incompressible Fluids, Reduced Order Models & Coarse Graining Techniques, Introduction to Machine Learning, Numerical Methods for the Solution of PDEs

MS in Computational Science & Engineering — Applied Mechanics

Thesis: CFD Analysis of Cold Storage for Performance Enhancement

Advisors: Ammar Mushtaq, Salma Sherbaz

Courses: Computing for CSE, Machine Learning, Computational Linear Algebra & Optimization, CFD, Boundary Layer Theory

MSc in Mathematics — Applied Mathematics

Thesis: Optimal Homotopy Asymptotic Method (OHAM) & Homotopy Perturbation Method (HPM)

Advisor: Hassan Khan

Courses: Advanced Calculus, Linear Algebra, Numerical Methods & Analysis, ODEs & PDEs, Mathematical Statistics, Optimization Theory

BSc in Mathematics & Physics — Applied Mathematics & Physics

Courses: Mathematics, Physics

IMT School for Advanced Studies Lucca, Lucca, Italy — Multi-Scale Analysis of Materials (MUSAM) Laboratory

I develop and validate Reduced-Order Models for nonlinear Fluid–Structure Interaction systems using high-fidelity finite volume simulations. My research also integrates Physics-Informed Neural Networks to enforce governing equations, improve predictive accuracy, and reduce computational cost.

Sant'Anna School of Advanced Studies — SMART Lab

Conducting research in Scientific Machine Learning, focusing on reduced-order modeling, surrogate modeling, and Physics-Informed Neural Networks (PINNs) for computational physics applications. Utilizing ITHACA-FV and PINA to develop efficient data-driven and physics-based modeling frameworks.

NUST School of Interdisciplinary Engineering & Sciences, Pakistan

Worked on numerical modelling and simulation of fluid dynamics problems. Contributed to the development of computational models, supported the communication of research outcomes through written reports and oral presentations, and assisted in the preparation of grant proposals.

DenseFusion, Pakistan

Contributed to projects in High-Performance Computing (HPC) and Computational Fluid Dynamics (CFD), with experience in MPI-based parallel programming, FEM/FVM derivations of the Navier–Stokes equations, and CFD simulations using ANSYS Fluent.

NUST School of Electrical Engineering & Computer Science, Pakistan

Supported the delivery of Numerical Methods for Engineering (undergraduate, 51 students) and Mathematical Methods of Artificial Intelligence (master's, 56 students) with course instructor Rai Sajjad Saif. Prepared and delivered lecture materials, graded quizzes and assignments, managed examination procedures, and responded to student inquiries.

Amal Academy

Selected for the Amal Academy Fellowship 2024 from a competitive pool of over 4,000 applicants. Completed a fully online program focused on communication, leadership, teamwork, and professional development.

Khyber Pakhtunkhwa Public Service Commission, Pakistan

Taught intermediate-level Mathematics, including complex numbers, integration, matrices, functions, and derivatives. Supported student success through engaging teaching methods and by making mathematical concepts more accessible, clear, and practical.

Reduced-Order Modeling of Turbulent Navier–Stokes Flow

Developed a reduced-order modeling framework for unsteady turbulent Navier–Stokes flow using Proper Orthogonal Decomposition with radial basis function (RBF) interpolation of eddy viscosity fields. Constructed snapshot databases for parametrized inlet velocity conditions and extracted dominant flow modes for velocity, pressure, and turbulence quantities. Integrated Galerkin projection with offline–online decomposition to enable efficient reduced-order prediction and reconstruction of turbulent flow dynamics.

POD · RBF · Galerkin Projection · OpenFOAM

Reduced-Order Simulation of Parametrized Viscous Flow Past a Cylinder

Developed a reduced-order model for unsteady Navier–Stokes flow around a circular cylinder with parametrized viscosity using OpenFOAM and ITHACA-FV. Applied Proper Orthogonal Decomposition with Galerkin projection and offline–online decomposition for efficient transient flow prediction and reconstruction across varying Reynolds numbers.

OpenFOAM · ITHACA-FV · POD · Unsteady NS

Non-Intrusive Reduced-Order Modeling Using POD and ML

A data-driven reduced-order modeling framework for parametric Navier–Stokes equations using Proper Orthogonal Decomposition (POD) with Radial Basis Function (RBF) and neural network regression. Constructed low-dimensional latent representations from high-fidelity CFD snapshot data and learned surrogate mappings from parameter space to reduced coefficients for fast inference. Evaluated POD–RBF and POD–NN models, demonstrating efficient and accurate flow field prediction across unseen inlet conditions.

POD–NN · POD–RBF · PyTorch · Surrogate Modeling

Reduced-Order Simulation of Flow in a Backward-Facing Step

A reduced-order model for parametrized steady Navier–Stokes flow over a backward-facing step using OpenFOAM and ITHACA-FV. Applied Proper Orthogonal Decomposition to snapshot data with varying viscosity to extract dominant flow modes and construct a reduced basis. Implemented Galerkin projection with offline–online decomposition for efficient parametric flow prediction and reconstruction.

OpenFOAM · ITHACA-FV · POD · Steady NS

Efficient Simulation of Parametric Heat Conduction via ROM

Developed a reduced-order modeling framework for a parametrized steady-state heat transfer problem using OpenFOAM and ITHACA-FV. Solution snapshots were generated for varying thermal diffusivity values to construct a low-dimensional representation of the system dynamics. The reduced model was obtained via Galerkin projection and offline–online decomposition, enabling efficient predictions while preserving full-order CFD accuracy.

OpenFOAM · ITHACA-FV · Heat Transfer

POD-Based Reduced Order Modeling of Lid-Driven Cavity Flow

Applied Proper Orthogonal Decomposition (POD) to a lid-driven cavity CFD problem solved with OpenFOAM's icoFoam solver. Extracted dominant spatial modes for velocity and pressure fields using the ITHACA-FV framework, and configured snapshot-based POD parameters including field selection, mode count, and time window via the ITHACAPODdict interface.

icoFoam · ITHACA-FV · POD

NUST Digital Twin through Advanced Modeling & Simulation

Conducted a CFD analysis of NUST H-12, Islamabad, to study airflow patterns and thermal gradients in the surrounding environment. The study provided insights for improving building placement, enhancing ventilation, reducing pollution, and mitigating the urban heat island effect, highlighting the role of strategic urban planning in thermal comfort and sustainable development.

Urban CFD · Digital Twin · Thermal Comfort

Machine Learning–Based Pipeline Leak Detection

Simulates the detection of leakage in pipelines and predicts leak characteristics using a Random Forest regression model. The workflow includes data extraction, preprocessing, and splitting into training and testing sets. The model estimates leak percentage based on location and type, and its performance is evaluated using mean absolute error before predicting leak characteristics for unseen pipeline data.

Random Forest · scikit-learn · Regression

Automated Fault Detection Using Random Forest Classification

System fault detection automated with a Random Forest Classifier. Fault-related data is extracted, preprocessed, and split for training and testing. The classifier is trained and evaluated for accuracy, enabling efficient fault identification and individual fault prediction for test samples — streamlining the detection process for quicker issue resolution.

Random Forest · Classification · Python

CO₂ Emissions by Vehicles — Multilinear Regression

A study of CO₂ emissions by vehicles using multilinear regression on data sourced from the Canadian government's official portal, focusing on prominent automobile manufacturers. Through rigorous analysis and model development, the predictive accuracy of the models was assessed, contributing to a comprehensive evaluation of their environmental impact.

Multilinear Regression · pandas · Open Data

Exploratory Data Analysis of Book Bestsellers

Exploratory Data Analysis on a book bestsellers dataset: importing and exploring structure, then analysing key variables such as genre, reviews, price, and user rating. Identifies top authors and showcases top-rated, most-reviewed, and highest-priced books. Calculates correlations and conducts a statistical test comparing fiction and non-fiction bestsellers, providing insight into trends and distributions.

pandas · EDA · Hypothesis Testing