CUDA matrix multiplication benchmarking on Jetson Orin Nano. Four implementations, three power modes, five matrix sizes. 99.5% mathematical validation. C++/CUDA and Python.

Standalone LLM inference benchmarking pipelines on AMD GPUs using ROCm, vLLM, MAD, and data visualization scripts.

🔍 Analyze CUDA matrix multiplication performance and power consumption on NVIDIA Jetson Orin Nano across multiple implementations and settings.

One-shot script to audit GPU, CUDA, PyTorch, CPU, and disk performance before debugging a slow or broken ML environment.

benchHUB is a Python-based project to parse, aggregate, and visualize system and performance benchmarks. It includes a Streamlit dashboard to display and compare results.

GPT-2 (124M) fixed-work distributed training benchmark on NYU BigPurple (Slurm) scaling 1→8× V100 across 2 nodes using DeepSpeed ZeRO-1 + FP16/AMP. Built a reproducible harness that writes training_metrics.json + RUN_COMPLETE.txt + launcher metadata per run, plus NCCL topology/log artifacts and Nsight Systems traces/summaries (NVTX + NCCL ranges).