GCP / TPU / Inference / Infra notes
Notes from Cloud AI Infra work — TPU v7 Ironwood, GPU inference (vLLM / SGLang), large-scale training pipelines, and the cloud infrastructure underneath. Written for engineers who want concrete details over abstract overviews.
This is the first post on blog.higcp.com. The blog is built with Jekyll on GitHub Pages, with a custom skin mimicking Google Cloud Console design: white background, Google Sans ...