HPC GPU & CUDA MPI Systems Engineer

Aditya Gowari

M.Sc. Scientific Computing HPC & GPU Systems

Pune, India +91-776**-***** ****************@*****.*** LinkedIn: linkedin.com/in/aditya-gowari GitHub: github.com/patzer-adi LeetCode: leetcode.com/u/patzer_adi_22

SUMMARY

M.Sc. Scientific Computing student specializing in HPC, GPU computing, and computational cryptography. Experi- enced in CUDA/MPI parallel system design, numerical algorithm development, and large-scale distributed compute on multi-node clusters. Strong fundamentals in C/C++, linear algebra, and performance optimization. TECHNICAL SKILLS

Languages: C/C++, Python

HPC / Parallel Computing: CUDA, MPI, OpenMP, Multithreading, Kernel & GPU Memory Optimization Libraries / Scripting: NTL, GMP, NumPy, SciPy, Matplotlib, SageMath, Bash Dev Tools: Git, CMake, GTest, Valgrind, SSH, LaTeX, Jupyter, VSCode Creative Tools: DaVinci Resolve, Canva

Systems: Linux (Arch, Ubuntu, Debian), macOS, Windows PROJECTS

1) Efficient Algorithms for the Zero Minor Problem Aug 2025 -- Present GitHub: github.com/patzer-adi/zero_minor_problem

Research project on detecting zero-determinant minors in n nECDLPkernelmatrices--- a structural problem at the intersection of linear algebra and computational cryptography. Team of 4. Tech Stack: C++, NTL, GMP, CUDA, MPI, CMake, Python, NumPy, Matplotlib, PBS, SSH, Linux Responsibilities:

• Built a CUDA APM kernel where each thread extracts a k k submatrix, runs modular Gaussian elimination (det mod p), and signals zero results via atomicOr; achieved an 81% detection rate on tested matrices, scaling to encryption level 40 on NVIDIA V100 (PARAM Brahma, IISER Pune), RTX 2080 Ti, and 750 Ti hardware over 150+ hours of compute (including a single 70-hour run).

• Distributed matrix generation across a 48-core MPI cluster over 300+ hours of parallel compute, producing 1000+ structurally valid ECDLP matrices; helped identify the minimum deviation level at which zero APM minors appear consistently.

• Patched the lasvegas-ecdlp C++/MPI codebase to generate matrices at scaled offset 0.2 --- fixed offset type and numberOfBits computation, enabling the antioffset dataset --- then helped analyze and visualize results using Python, NumPy, and Matplotlib.

2) Numerical Computing Library in C++ Jan 2026 -- Present GitHub: github.com/patzer-adi/Numerical_Computing

C++ numerical methods library with runtime GPU/CPU dispatch --- algorithms execute on CUDA when a GPU is available, falling back to CPU automatically.

Tech Stack: C++, CUDA, CMake, GTest, Valgrind, Linux (Arch, Debian) Responsibilities:

• Implemented Gaussian Elimination, Gauss-Jordan, LU Decomposition (Doolittle, Crout, Cholesky), Gauss-Jacobi, and Gauss-Seidel; validated on matrices up to 225 225 across CPU and CUDA.

• Built a Complex class with full arithmetic and root-finding solvers: Newton-Raphson, Fixed-Point Iteration, and Bisection Method.

• Applied OOP patterns throughout: virtual functions, operator overloading, friend functions, and file-based matrix I/O.

EDUCATION

Master of Science in Scientific Computing 2025 -- 2027 SCMS, Savitribai Phule Pune University, Pune (Pursuing) Bachelor of Science in Computer Science 2022 -- 2025 MES Abasaheb Garware College, Savitribai Phule Pune University, Pune CGPA: 9.03/10 EXTRACURRICULAR ACTIVITIES

• Helped edit and produce a podcast video using DaVinci Resolve and Canva, featured on the School of Scientific Computing YouTube channel.

• Volunteered to help host an alumni meet with around 100 guests in August 2025.

• Helped organize and took part in a chess tournament during undergraduate studies.

Contact this candidate