Projects

Selected work in GPU performance characterization, compiler validation, and CI/CD for HPC. Each project is summarized by what shipped and what was measured — not the methodology behind it. Links go straight to source, papers, and dashboards.

Research Projects

OpenACC Validation and Verification (V&V) Testsuite

Active Development

Leading the development of a comprehensive validation and verification testsuite for the OpenACC Programming Model. This project ensures compiler compliance and reliability across diverse GPU architectures, supporting the broader HPC community.

OpenACC 3.0+

spec coverage

4 vendors

NVIDIA HPC SDK, GCC, Clang/LLVM, Cray

3 DOE systems

Summit, Frontier, Perlmutter

Best Poster — SC22

ACM SRC, DARWIN '23

OpenACCCUDAGPU ComputingCompiler TestingC/C++

GitHub Repository Project Website SC24-W WACCPD 2022

Stewardship for Programming Systems and Tools (S4PST)

Production

Contributing to a predictive ecosystem for HPC software sustainability. Our CI/CD pipeline for LLVM Clang and new-Flang's OpenMP Offloading implementation runs comprehensive test suites and benchmarks on cutting-edge GPU hardware including NVIDIA H100, GH200, and AMD MI210/MI300A.

4 architectures

H100, GH200, MI210, MI300A

15+ apps

DOE exascale workloads validated

−60%

manual testing overhead

IWOMP '24

published, peer reviewed

OpenMPLLVMCI/CDDockerGitLabHPCGPU Benchmarking

CI/CD Pipeline S4PST Website arXiv IWOMP 2024

LLM4VV: AI-Driven Compiler Testing

Research Phase

Developing novel approaches to leverage Large Language Models for automated test generation and validation in compiler verification workflows. This research explores the intersection of AI and systems software to improve testing efficiency and coverage.

2 papers

FGCS journal + SC24-W workshop

GPT-class

LLM-as-judge for compiler tests

OpenACC + OpenMP

directive coverage

LLMsGPT ModelsAutomated TestingCompiler ValidationPythonAI/ML

arXiv Paper LLM-as-a-Judge Paper SC24-W FGCS

OpenMP Offloading CI/CD Framework

Active Development

Comprehensive continuous integration and deployment framework for validating OpenMP GPU offloading implementations. Focuses on automated testing, performance benchmarking, and cross-platform compatibility across diverse GPU architectures.

IWOMP '24

20th Intl. Workshop on OpenMP

GitLab CI

automated H100/GH200/MI300A runs

LLVM Clang + Flang

compilers under test

OpenMPCI/CDGPU OffloadingPerformance TestingLLVMClang

Research Paper IWOMP 2024

Multi-Architecture GPU Validation Framework

Active Development

Developing unified testing methodologies for GPU computing across NVIDIA (H100, GH200) and AMD (MI300A) architectures. Ensures consistent performance and reliability for HPC applications across diverse hardware platforms.

FP8, 2:4 sparsity

characterized on MI300A

5th-gen TC

Blackwell FP4/FP6 microbenchmarks

IPDPS '26

accepted

CUDAHIPROCmGPU ComputingPerformance AnalysisCross-Platform Testing

MI300A FP8 & Sparsity (arXiv) arXiv IWOMP 2024

Side Projects & Coursework

Self-contained projects from coursework, hackathons, and exploratory work.

Quantum Error Correction — Surface Code Study

Quantum

Course-driven deep dive on stabilizer codes and the surface code, capped with a written report and a 30-page technical write-up.

30+ pages

technical writeup

Surface code

decoder analysis

Report (PDF) Primer (PDF) Blog post

MARS — Multi-Architecture Roofline Study

Performance Analysis

Roofline-model performance characterization of GEMM and FDTD on CPU+GPU. Measured cache-hit behavior and bandwidth-bound vs compute-bound regimes across problem sizes.

GEMM + FDTD

kernels measured

CPU + GPU

rooflines compared

Report (PDF) Poster (PDF) Slides (PDF) Blog post

HenStreet Hacks 2025 — SallieMae Track

Hackathon

Built a hackathon prototype in a fast-turnaround team setting for the SallieMae industry track at the University of Delaware.

48-hour

build window

Team of 4

cross-functional

Presentation (PDF) Blog post

Looking for code? Most of my open-source work lives on GitHub and in the OpenACCV-V repository.