Designing Machine Learning Systems

I’ll admit: I rolled my eyes at first. Another ML book? But *Designing Machine Learning Systems* hooked me by page three—not with breakthrough algorithms, but with Huyen’s brutal honesty about the *social engineering* required to ship ML. Her chapter on "The Hidden Cost of Iteration" should be required reading for every PM who’s ever demanded ‘just one more A/B test.’ The audiobook’s strength lies in how Kathleen Li delivers Huyen’s dry humor and no-BS asides. When Huyen skewers the ‘move fast and break things’ mentality in ML (‘*You can’t “break” a model and just roll back—your users’ trust is gone*’), Li’s timing makes it land like a punchline. That said, the audiobook isn’t perfect. The middle section on monitoring drags slightly—Huyen’s repetition of ‘*you must log everything*’ feels overemphasized, and Li’s cadence here flattens into a checklist monotone. I also wished for more concrete examples in the ‘Ethics’ chapter; Huyen names biases but stops short of prescribing actionable guardrails, leaving listeners with a ‘now what?’ hangover. Still, the final chapters on *decommissioning* models (yes, *killing* them) are gold, especially Li’s delivery of Huyen’s refrain: ‘*The most responsible ML is the ML you don’t build.*’ For engineers tired of theoretical fluff or managers drowning in ML hype, this is the audiobook that’ll make you nod, wince, and—crucially—*change how you work*.