Long before engineering existed, humans watched fire spread, water find its path, smoke rise and heat move from one object to another. Our instinctive understanding of flow and heat began with nature itself.

Today, I sit in front of a simulation and try to persuade a computer to reproduce those same behaviours. We describe heat using Fourier's 200-year-old mathematics and fluid motion using the Navier-Stokes equations—born in the nineteenth century, implemented through decades of numerical methods and executed on modern hardware.

Now AI has joined the conversation: friend, enemy or an extremely confident assistant? Perhaps all three.

It can help us code faster and explore more possibilities, but it cannot negotiate with nature. The equations must converge, the results must agree with physics, and engineering judgment still gets the final vote.

To help transportation move people and the things humanity depends on more safely, efficiently and sustainably—using thermal-fluid engineering, combustion, electrification, simulation automation and physics-aware AI.

• Battery & E-Powertrain — thermal management, thermal runaway , electro-thermal hotspots and component level system simulation
• Combustion & Aftertreatment — SI/CI combustion, RCCI , charge motion, spray behaviour, EATS
• Thermal & Flow — coolant flow, conjugate heat transfer, temperature fields, DOE , Optimisation and test correlation
• Component-to-System Simulation — connecting detailed CFD with cooling circuits, controls and vehicle-level behaviour
• Engineering AI — reduced-order models, surrogates, explainability and optimization without losing the physics

• In-cylinder PIV flow reconstruction with POD/PCA
• Explainable combustion ML with Random Forest, XGBoost and SHAP
• Thermal surrogate modelling, constrained optimization and an engineering dashboard
• Vehicle-aerodynamics ML using real DrivAerNet++ design and CFD data

Physics first. Validation before claims. Fancy models still need to obey thermodynamics.

Industrial simulation: STAR-CCM+ · CONVERGE · ANSA · GT-SUITE · Ansys . Modelica

Scientific computing: Python · Wolfram · Matlab · Openfoam

Applied engineering ML: Scikit-learn · Pytorch · RAG · Surrogate Modelling . PINNS

Reproducible delivery: Git · GitHub · Kaggle · Codespaces · Streamlit

• EV integrated thermal-management and 1D thermo-fluid system models
• OpenModelica/Python workflows with concepts transferable to GT-SUITE, AMESim and Simulink
• Component maps connecting 3D CFD to vehicle-level system simulation
• Controls, calibration, uncertainty and independent validation
• PyTorch for thermal-field and engineering deep learning
• CNN/U-Net thermal hotspot prediction
• Physics-informed and hybrid first-principles/ML models
• Neural operators, digital twins and uncertainty-aware surrogates

Always happy to exchange ideas about CFD, engine CHT, combustion, battery thermal safety, system simulation, automation or Scientific ML.

LinkedIn · Engineering-AI Portfolio

_{My curiosity is currently unlimited. My GPU budget has a much stricter boundary condition.}