stas00/ml-engineering: An opinionated overview of ML and AI engineering realities

github.com21 min read

The author presents an opinionated, practical overview of machine learning engineering realities, focusing on hardware economics, infrastructure bottlenecks, compute procurement options, and software maintenance challenges. The guide details the mathematics behind calculating training costs and MFU, evaluates various hardware accelerators, and discusses strategies for managing storage, networking, and vendor lock-in. Open article →

More on this topic