Neo Kim shares 15 system‑design cases
- Neo Kim resurfaced a 15-case system-design study list on February 27, bundling explainers for ChatGPT, Google Search, YouTube, Kafka, S3, and payments. - The list spans 15 named systems, from Uber ETA and Twitter timelines to stock exchanges, mixing consumer apps with infrastructure and core backend patterns. - It matters because interview prep is shifting from toy prompts toward real production systems, where tradeoffs and scaling choices are easier to practice concretely.
System-design interview prep is usually too abstract. You get a prompt like “design Dropbox” or “design Twitter,” then you’re supposed to magically know which tradeoffs matter. Neo Kim’s latest post goes the other way. On February 27, he shared a 15-case reading list that turns interview prep into concrete architecture drills built around real systems like ChatGPT, Google Search, Kafka, YouTube, Amazon S3, and payment systems. (substack.com) ### What actually got shared? The post is basically a curated index. Kim linked 15 case studies from his System Design Newsletter and site, pitching them as the set to learn if you want to get good at system design. The named cases are: ChatGPT, Google Search, Uber ETA, Amazon S3, YouTube, Kafka, WhatsApp, Spotify, Slack, Reddit, Bluesky, Twitter timeline, URL shortener, payment systems, and stock exchanges. (substack.c([substack.com)es that format work better? Because system design is rarely about memorizing one “correct” architecture. It’s about learning how different constraints force different shapes. A YouTube-style system teaches storage, transcoding, CDN strategy, and fan-out. Kafka teaches partitioning, ordering, durability, and consumer coordination. Google Search pushes you toward crawling, indexing, ranking, and freshness. Put si(substack.com) ### Why these 15? Turns out the list is broad on purpose. Some cases are user-facing apps — WhatsApp, Spotify, Reddit, Slack. Some are infrastructure primitives — Kafka, S3, URL shorteners. Some are high-stakes transactional systems — payments and stock exchanges. That spread matters because interviewers often want to see whether a candidate can move between latency-heavy, throughput-heavy, consistency-heavy, and cost-sensitive designs without getting stuck in one template. (substack.com) ### Is this new material? The post is new-ish, but the pieces are mostly drawn from Kim’s existing newsletter archive and systemdesign.one library. So the news is less “15 brand-new essays dropped today” and more “here is the compact syllabus.” That makes the thread useful as a map. Instead of digging through a long archive, readers get one entry point into the most interview-friendly examples. (substack.com)ctually practice with these? Three things, mainly. First, requirement scoping — what are the must-haves, what can be cut, and what scale are we designing for? Second, component selection — caches, queues, databases, indexes, CDNs, stream processors, and so on. Third, tradeoff talk — why you picked availability over strict consistency, or batching over immediate processing, or precomputation over o(substack.com)re diagram matters, but the explanation matters more. (softarchi.blog) ### Why are real systems better than toy prompts? Because they give you a mental anchor. “Design a payment system” instantly raises fraud, idempotency, retries, ledger correctness, and reconciliation. “Design Kafka” makes you think about logs, partitions, brokers, and replay. It’s like lifting with real weights instead of waving your arms around — the constraints are tangible, so your reasoning g(softarchi.blog)ea that system-design prep is becoming a library game, not a flash-card game. (softarchi.blog) ### Is this aimed only at FAANG interviews? Mostly, but not only. Kim frames his work around helping engineers get good at system design, and his newsletter has grown into a large audience product with hundreds of thousands of subscribers. That tells you something about demand — engineers are using these breakdowns both for interviews and for general architecture literacy. The same case study tha(softarchi.blog)ng decision on the job. (newsletter.systemdesign.one) ### So what’s the takeaway? The useful part of this post is not the number 15. It’s the framing. Kim turned system-design prep from vague advice into a concrete reading sequence: study real systems, compare the constraints, and practice defending tradeoffs out loud. That is much closer to how strong interview answers — and real engineering decisions — actually work. (substack.com)
