Phase 11 · Complete System Designs
End-to-end case studies from classic to advanced.
- The Recurring Patterns: Recognise the Problem Shape, Pick the Solution Ladder: Seven problem shapes cover most design prompts. Learn the phrases that reveal each one, the ladder of solutions from simple to advanced, the numbers at which each rung breaks, and where the app already works each pattern end to end.
- Classic Designs: URL Shortener, Pastebin, KV Store, Unique IDs & Notifications: The warm-up designs every interviewer reaches for — walked end-to-end with the 4-step framework, real numbers, and the deep dives that separate senior from mid-level.
- Location Systems: Proximity Service, Nearby Friends, Google Maps & Uber: Geospatial indexing, real-time location fan-out, map tiles and routing, and ride-hailing dispatch — the designs where "where is it?" meets "how fast can I answer?".
- Social & Media: Chat/WhatsApp, News Feed (Twitter/Instagram) & YouTube/Netflix: Real-time messaging with ordering and receipts, feeds that survive celebrities, and video pipelines that transcode, stream and count at planetary scale.
- Infrastructure Systems: Distributed Message Queue, Metrics Monitoring, Ad Click Aggregation, Job Scheduler & Logging Platform: Designing the systems other systems stand on — a Kafka-like log, a time-series monitoring stack, an exactly-once click pipeline, a distributed scheduler, and a log platform with tiered storage.
- Commerce & Money: Hotel/Ticket Booking, E-commerce, Payment System & Digital Wallet: Five end-to-end case studies where a wrong answer costs real money: inventory that must never oversell, seats held under flash-sale load, carts that turn into orders, payments that must happen exactly once, and wallets that must balance to the paisa.
- Storage Systems: Dropbox/Google Drive, Distributed Email & Distributed File System: Three case studies about keeping user bytes safe at petabyte scale: file sync with block-level dedup and delta sync, a Gmail-scale email service, and a GFS/HDFS-style distributed file system with leases and replication.
- Advanced: Leaderboard, Stock Exchange, Distributed Cache, Global Rate Limiter, Search Engine/Typeahead, Ad Serving, Recommendation System & Multi-tenant SaaS: Eight senior/staff-level case studies, each crisp but deep: Redis sorted-set leaderboards, a microsecond matching engine, a sharded cache tier, rate limiting across regions, typeahead and web search, real-time ad auctions, multi-stage recommenders, and tenant isolation in SaaS.
- Real-time, Media & AI: Live Streaming (Hotstar/Twitch), Video Conferencing (Zoom/Meet), ChatGPT-style LLM Service & Google Docs: Four designs interviewers added to their rotation in 2024–2026: one-to-50-million video fan-out, sub-150 ms group calls, GPU-bound token streaming, and convergent real-time co-editing.
- More Classics: Web Crawler, Top-K / Trending, Distributed Counters (Likes & Views), Online Judge (LeetCode) & Online Auction (eBay): Five designs that interviewers keep asking and that test different muscles: politeness and dedup at web scale, streaming heavy hitters with sketches, counting under extreme write contention, running untrusted code safely during a contest spike, and strongly consistent bidding against a deadline.
- Consumer & Marketplace: Flash Sale (Big Billion Days), Price Tracker & Alerts (CamelCamelCamel), Tinder, News Aggregator (Google News) & Strava: Five consumer designs that each hide one hard problem: selling 50K units to 5M people in a minute without overselling or rewarding bots, matching millions of price rules without spamming anyone, detecting a mutual like exactly once, turning 3M articles a day into a few hundred stories, and timing an athlete over a GPS trace that never lies on the segment.
- Platform, Fintech & AI: Robinhood-style Brokerage, CI/CD Runner (GitHub Actions), Online Chess Server, Calendar & Scheduling (Google Calendar / Calendly) & RAG Document Q&A: Five designs that interviewers moved into their loops in 2025–2026: a retail brokerage that must stay correct at the opening bell, a CI platform that runs strangers' code fairly and safely, a real-time game server where 300 ms decides a game, a calendar where time itself is the hard part, and an enterprise "chat with your docs" system where retrieval, not the LLM, decides quality.
- Instagram & Food Delivery (Swiggy / Zomato / DoorDash): Two of the most-asked product designs, taken past the generic building blocks: Instagram's media pipeline, ranked feed, Stories, Explore and celebrity-scale engagement; and a three-sided food-delivery marketplace with serviceability, an idempotent order state machine, dispatch with batching, ETA promises, delivery fees and peak-hour failure handling.
- AI Agents & LLM Apps: Agent Loops, Tools/MCP, Agentic RAG, Security, Evals & Troubleshooting: The 2026 "AI system design" round: build agents as distributed systems with budgets, validated tools, durable steps, least privilege and evals, then debug them when they are slow, wrong, stuck or expensive.