Documenting my path through Computer Science at ASU—projects, learnings, and insights along the way.
I pointed autoresearch at an AMD Radeon AI PRO R9700 and let it run 316 experiments. The biggest win wasn't architecture — it was enabling torch.compile on RDNA4.
A personal build log covering hardware selection, assembly, and getting local AI inference running on RDNA 4 with ROCm. 32GB of VRAM, no cloud bills.
I gave Claude Code an H100 GPU, Karpathy's autoresearch framework, and a full night to autonomously optimize a language model. Over 100 experiments later — 16 of which passed — here's the full breakdown.
Introducing my new blog where I'll be documenting my experiences as a Computer Science graduate student at ASU, including projects, coursework insights, and industry learnings.