MSc, Computational Science
American University of Beirut. Coursework includes optimization, GPU computing, statistical learning, finite elements, and vortex methods.
CV / computational science / physical oceanography
Computational science MSc researcher at the American University of Beirut, with work across geophysical fluid dynamics, vortex methods, MITgcm, GPU computing, data assimilation, and LLM/RAG systems.
Research interests: computational physics, physical oceanography, numerical modeling, parallel computing, scientific ML, and interfaces that make advanced simulation tools easier to use.
Current work applies numerical frameworks to vorticity dynamics, energy transfer, and potential-vorticity conservation in quasi-geostrophic shallow-water systems.
Training in computational science, theoretical physics, pure mathematics, and scientific computing.
American University of Beirut. Coursework includes optimization, GPU computing, statistical learning, finite elements, and vortex methods.
American University of Beirut. Focus on analysis, differential equations, algebra, manifolds, probability, and numerical computing.
Pitzer College, Claremont, California. Statistical physics, French, and R programming.
Current research connects vortex methods, ocean dynamics, and production numerical models.
Building code and workflows to simulate figures and dynamics from Marshall's atmosphere-ocean climate dynamics text.
Developing a computational model for quasi-geostrophic shallow-water equations over spherical geometry.
Microflows and Microscale Heat Transfer Laboratory, American University of Beirut.
Computational work spanning LLM agents, GPU algorithms, PDEs, optimization, and data assimilation.
LLM/RAG interface for querying documentation, modifying code, and executing simulations through natural language.
Parallel implementation using CUDA and Octopus HPC, with reduction trees, stencils, coarsening, and occupancy optimization.
Finite-element and Galerkin analysis using FreeFem and MATLAB, with numerical and analytical comparison.
Data assimilation for the Van der Pol oscillator using adjoint-based BFGS and Bayesian optimization.
Original CV file copied from the local source document.