About my work
I work at the intersection of machine learning and systems, focusing on GPU performance, CUDA kernels, PTX-level optimization, and training and inference efficiency at scale.
Current Obsession
Lately I have been spending a lot of time tuning programs around how the GPU actually behaves. That has taken me deeper into CUDA, kernels, PTX, SASS, and the low-level details behind training and inference performance.
How I Like To Work
I like problems without clean, predefined solutions. The kind that force me to read, benchmark, experiment, and piece things together on my own until the answer becomes clear.