Hi, I’m Joseph.

This is Joseph’s Notebook — a place to document ideas, experiments, and learnings. I write about deep learning, LLMs, AI/ML infrastructure, and related topics. Occasionally you’ll find notes on SRE, distributed systems, and other tech that catches my interest.

Interests

  • Large language models and transformers
  • RAG, embeddings, and retrieval systems
  • Model training, quantization, and inference optimization
  • Coding agents and AI tooling
  • GitHub — source and discussions
  • Blog content is driven by GitHub Issues — feel free to open one for discussion

About this blog

This site is forked from the blog of ninehills（九原客）, an AI Infra practitioner I came across on X/Twitter whose work background aligns closely with mine. I’m a fan of his writing on LLM training and his minimal GitHub Issues–driven blog setup. I kept his posts in the deep-learning / LLM / GPU / inference space, and built on that foundation with several improvements: configurability, shadcn/ui-inspired styling, CJK font rendering, full-text search, LaTeX math, Mermaid diagrams, syntax highlighting, on-demand JavaScript loading (MathJax, Mermaid, and Fuse.js are only loaded when needed), and related tooling. Articles whose original author is ninehills carry a short repost notice at the top and link back to his GitHub Issues blog.

“Nine hills” — a small corner of the web for scattered thoughts. This site continues that spirit.