Aamir S.

Your research depends on a computation that will not cooperate. A solver that does not converge, a method from a paper that needs implementing, or a simulation too slow to use. I build that in Python and PyTorch, from the equations up. I am a PhD researcher in applied mathematics at GIK Institute, Top Rated with a 100% Job Success Score. I write the numerical method myself rather than wrapping a package. WHAT I HAVE DELIVERED For a number theory researcher I built seven computational programs over several months, including spectral analysis tools and a geometry engine, and packaged them as a desktop application for Windows and macOS. In the client's words, I "took interest in the overall goals and intentions for the project." For a contact mechanics study I built an inverse physics-informed neural network in PyTorch that recovered the contact radius to 99.7% of the analytical value. For a turbulence project I adapted a PINN with sparse regression to recover the governing Navier-Stokes terms for 2D Kolmogorov flow at Re = 10,000. For a small business I built a shift settlement tool. It tracks stock passed between four workers and works out what each one owes, with the history saved behind a login. RESEARCH My paper on physics-informed neural networks for the regularized long-wave equation is on arXiv (2511.11638). A second paper, on the coupled Klein-Gordon-Schrodinger system, is under review. It reports errors 2.6 to 5.8 times lower than a published radial basis function method. I am on a fully funded HEC-IRSIP research visit at Imperial College London from September 2026 to March 2027. I have also taught Calculus and Numerical Analysis at GIKI. WHAT I TAKE ON I take on custom solvers and simulations in Python or MATLAB, implementations of methods from papers, and surrogate models for slow CFD or FEM runs. I also debug scientific and machine learning code that trains but gives the wrong answer, and turn calculations a business does by hand into a working tool. For AI evaluation teams, I write and review research-level scientific coding problems with Python reference solutions and hidden tests. I do not take coursework or tutoring, and I do not use ANSYS or COMSOL. Describe your problem in a message. I will tell you plainly whether I can solve it and how. Scientific Computing | Numerical Methods | Partial Differential Equations | PDE Solver | Physics-Informed Neural Networks | PINN | Inverse Problems | Surrogate Modeling | Simulation | Mathematical Modeling | Algorithm Development | Python | NumPy | SciPy | PyTorch | DeepXDE | MATLAB | Research Scientist