I am an Engineering Mechanics undergraduate working on GPU numerical linear algebra, scientific computing, and open-source systems.

My recent work focuses on CUDA performance engineering, sparse matrix operations, CFD linear algebra infrastructure, and graphics-pipeline debugging, especially where numerical methods meet low-level system behavior.

• GPU performance engineering — CUDA data movement, sparse matrix-vector products, batched GEMM paths, and solver-side performance.
• Sparse solver infrastructure — cuSPARSE, cuBLAS, Krylov-solver data flow, and CFD linear algebra paths.
• Solver correctness and validation — block-sparse matrix operations, minimal numerical counterexamples, and reproducible correctness tests.
• Scientific computing — numerical methods, nonlinear mechanics, FEM validation, and simulation reliability.

• SU2 CUDA linear algebra path
  Reducing redundant Jacobian uploads during linear solves and investigating CUDA block-sparse matvec correctness in the GPU solver path.

• CUDA sparse solver optimization
  Reusing cuSPARSE SpMV preprocessing in amgcl's CUDA CSR backend to reduce repeated CSR partition/preprocessing overhead in iterative solves.

• ArrayFire CUDA batched GEMM
  Adding a strided-batched GEMM fast path for compatible batch layouts to avoid pointer-array setup and host-to-device pointer copies.

• CUTLASS runtime datatype mapping
  Improving runtime datatype mapping paths in CUTLASS library tooling.

• Hyprland ICC / blur rendering investigation
  Debugging ICC-enabled blur transparency and color-pipeline interactions in the compositor render path.

• Nonlinear beam deflection computation
  Numerical calculation and FEM validation for the failure boundary of linear beam theory under large deflection.