🏗️ System Design

A repeatable 6-step structure, capacity math, latency numbers, CAP/PACELC, and consistency models so you never freeze on an open-ended design question.

Load balancers, gateways, caching, CDN, queues vs pub/sub vs streams, databases, object storage, search, rate limiters, service discovery — each with when to reach for it and the tradeoff.

The NoSQL families and when to use each, B-tree vs LSM indexing, ACID vs BASE and isolation levels, sharding & rebalancing, replication topologies, and access-patterns-drive-the-schema.

How Node stays fast on one thread — libuv loop phases, the thread pool, CPU vs IO-bound work, clustering & worker_threads, streams/backpressure, GC, and production pitfalls.

Two warm-up designs (URL shortener, rate limiter) plus the reusable patterns interviewers probe: fan-out, CQRS, event sourcing, idempotency, outbox, consistent hashing, replication, sharding, hot keys, backpressure.

A repeatable method for frontend design rounds: requirements → API/data → component architecture → state → rendering (CSR/SSR/SSG/ISR) → performance budget → network/caching → a11y → offline.

A deep concept walkthrough of the feed: fan-out on write vs read, the hybrid for celebrities, ranking, cursor pagination, caching, and the client-side concerns (virtualization, optimistic UI).

Caching layers, write strategies (aside/through/back), eviction, the three big failure modes (stampede, penetration, avalanche), invalidation & consistency, CDN internals, and Redis data structures.

API styles and when to choose each, REST done right (status codes, idempotency, pagination, versioning), real-time options, sync vs async, and resilience: retries, timeouts, circuit breakers, rate limiting.