I’m a Computational Engineer with a background in mechanical engineering, numerical simulation, and Scientific Machine Learning (SciML).
My work sits at the intersection of physics-based simulation, scientific computing, and machine learning, with a particular interest in using modern computational methods to make engineering simulation, design, and optimization faster and more scalable.
I work with CFD, numerical methods, HPC, deep learning, and engineering optimization, and I enjoy building computational tools that connect physical models with data-driven approaches.
- Computational Fluid Dynamics (CFD)
- Scientific Machine Learning (SciML)
- Physics-Informed Neural Networks (PINNs)
- Surrogate Modeling & Reduced-Order Modeling
- Generative Design & Shape Optimization
- Deep Learning for Scientific Computing
- High-Performance Computing (HPC)
- Numerical Methods & Scientific Computing
- Multiphysics Simulation
- AI-assisted Engineering Design
Developing machine learning approaches that incorporate physical knowledge into engineering problems, including:
- Physics-Informed Neural Networks for fluid mechanics and heat transfer
- Neural operators and surrogate models
- Learning-based acceleration of computational simulations
- Data-driven modeling of physical systems
Building numerical and simulation workflows for engineering applications:
- CFD using OpenFOAM, ANSYS Fluent and CFX
- Finite Volume and Finite Element methods
- Multiphase flow and transport phenomena
- Porous-media flow and evaporation
- Numerical linear algebra and iterative solvers
- Automated simulation workflows
Exploring how machine learning can improve computational design:
- Generative modeling of airfoils and turbine geometries
- Latent-space design optimization
- Surrogate-assisted optimization
- Aerodynamic shape generation
- ML-assisted turbine design workflows
| Project | Description |
|---|---|
| HPC CFD Automation Framework | Automated simulation, preprocessing, postprocessing and data workflows for CFD |
| LidDrivenCavityPINN | Physics-Informed Neural Network for incompressible lid-driven cavity flow |
| FVM Solver | Implementation of finite-volume numerical methods for computational fluid dynamics |
| Project | Description |
|---|---|
| PINNs from Scratch | Step-by-step implementations of Physics-Informed Neural Networks for engineering problems |
| Neural Operators | Exploration of operator-learning approaches for physical systems |
| Surrogate Models | Machine-learning models for accelerating computationally expensive simulations |
| Transformer From Scratch | Implementation of transformer architectures to understand modern deep learning from first principles |
| Project | Description |
|---|---|
| PyTorch-Bezier-GAN | Generative model for aerodynamic shape generation using Bézier parameterization |
| bezier-hingegan | GAN-based generation of diverse aerodynamic geometries using hinge-loss training |
| HydFoil_latent | Latent-space representation and exploration of hydrofoil geometries |
| LatentSpace-Optimization | Optimization of engineering designs directly in learned latent spaces |
| Axial Turbine Optimization | ML-assisted turbine design and optimization workflow |
Python · C/C++ · MATLAB · Bash · Git · Linux
PyTorch · TensorFlow · NumPy · SciPy · scikit-learn
PINNs · Neural Networks · GANs · VAEs · Transformers · Neural Operators · Surrogate Models
OpenFOAM · ANSYS Fluent · ANSYS CFX
Finite Volume Method (FVM) · Finite Element Method (FEM) · Navier–Stokes Equations · Multiphase Flow · Heat Transfer · Porous Media · Numerical Linear Algebra
MPI · OpenMP · Linux · HPC Clusters
Parallel computing · Simulation automation · Data pipelines · Large-scale numerical simulations
CATIA V5 · SolidWorks
Parametric modeling · DFM · GD&T · Engineering design · Computational geometry · Design optimization
I’m particularly interested in problems where traditional engineering simulation and modern machine learning can complement each other.
Rather than treating ML as a replacement for physics-based methods, I’m interested in questions such as:
How can we combine physical models, numerical methods, and machine learning to build better computational tools for engineering?
This includes using ML to accelerate expensive simulations, learning reduced representations of physical systems, embedding physical constraints into neural networks, and exploring new approaches to computational design.
I regularly write about CFD, Scientific Machine Learning, PINNs, numerical methods, and computational engineering, with a focus on making advanced topics accessible to engineers.
I am also developing a PINNs from Scratch tutorial series aimed at mechanical and computational engineers who want to learn how physics-informed machine learning works by implementing the methods themselves.
If you find my work on Scientific Machine Learning, computational engineering, and Physics-Informed Neural Networks useful, you can support my open-source projects and educational content at
I enjoy taking an engineering problem, starting from the governing equations, implementing the numerical method, and then asking:
"Can machine learning make this better?"
That's essentially where my interest in Computational Engineering + SciML comes from.