Gaurav Kumar

-Refactored a high-dimensional PDE solver from a GPU-based PyTorch architecture to a CPU-optimized SciPy backend, resolving critical Out-of-Memory (OOM) bottlenecks on cloud infrastructure. -Optimized numerical performance by developing vectorized evaluation routines for Chebyshev polynomial derivatives in NumPy. -Built a synthetic data generation pipeline to validate the convergence of sparse approximations against theoretical benchmarks on high-dimensional parameterized boundaries.