Li Bojie open-sources 'Understanding AI Infra' book, 7,500+ GitHub stars in weeks
udmrzn · x · 2026-11-14
Li Bojie has open-sourced a new book, Understanding AI Infra, which hit 7,500+ GitHub stars within weeks of its late-August repo creation. It's the companion to his 50,000+ star Understanding AI Agents.
- Inverted methodology: instead of listing frameworks and parameters, it derives designs from hardware constraints via five questions — what to move, how much, how many times, through where, and who must wait
- 12 chapters: model architecture, accelerators, supernodes, datacenter networking, inference optimization, distributed inference, training systems, and edge-cloud collaboration
- Companion calculator: runs on the Python standard library alone — no GPU or model weights needed, so you can recompute every number yourself
- Formats: community translations in English, Traditional Chinese and Russian; PDF, EPUB and web reading
Still a draft under active revision.
More from Infra
- Bernstein: Neoclouds earn ~6x more revenue per megawatt than Bitcoin miners — Beth_Kindig · 2026-11-14
- ZONOS3-Mini: MoE TTS model runs realtime on a Pixel 9 and MacBook Air — cocktailpeanut · 2026-11-14
- The Periodic Table of Inference: 12 LLM serving concepts as meme elements — sophiamyang · 2026-11-14
- Open-weights TTS model Tontaube V1.1 runs near real-time on a desktop CPU in 4-bit — EAVDR · 2026-11-14
- MiniMax H3 Image goes live on TensorScale, claimed 2.3x faster and 5x cheaper than fal at 4K — TerryTangYuan · 2026-11-14
- Prime Intellect jumps into top 10 third-party inference providers at 2.22T weekly tokens — Ronangmi · 2026-11-14