High-Level Design (HLD)
Architecture-level system design: building blocks, distributed systems theory, classic and large-scale case studies, AI system design, frontend system design, and reliability.
HLD is about the shape of the whole system: which services, which data stores, how requests flow, how it scales, and how it fails. The pages below run from beginner to advanced.
- 01Building Blocks
The components every HLD answer is made of: DNS, CDN, load balancers, API gateways, caches, SQL and NoSQL databases, replication, sharding, consistent hashing, queues and streams, object storage, search, and real-time protocols.
Beginner - 02Distributed Systems
Advanced HLD theory: unreliable networks and clocks, replication and quorums, consensus (Raft), distributed locks, isolation levels, 2PC, sagas, outbox, idempotency, resilience patterns, event-driven design, multi-region, and blast-radius control.
AdvancedTable of Contents1. Why Distributed Systems Are Hard2. Time and Ordering3. Replication in Depth4. Consensus and Raft5. Locks, Leases and Fencing6. Isolation Levels and Anomalies7. Distributed Transactions8. Idempotency and Exactly-Once9. Rate Limiting Algorithms10. Resilience Patterns11. Event-Driven Architecture+4 more - 03Classic Designs
Eight classic HLD case studies with requirements, estimates, APIs, data models, flowcharts and trade-offs: URL shortener, rate limiter, unique ID generator, key-value store, notification system, web crawler, typeahead, and distributed cache.
Beginner to mid - 04Large-Scale Designs
Twelve senior-level HLD case studies with estimates, flowcharts and deep dives: news feed, chat, video streaming, ride hailing, file sync, ticket booking, payments, ad click aggregation, job scheduler, leaderboard, collaborative editing, and a monitoring system.
Mid to senior - 05AI System Design
System design for LLM products: token economics, LLM inference serving (continuous batching, PagedAttention, prefix caching), a ChatGPT-style assistant, RAG with hybrid search and reranking, agents and MCP, AI gateways, evaluation, reliability and cost.
Mid to seniorTable of Contents1. What Is Different About AI Systems2. LLM Concepts a System Designer Needs3. LLM Inference Serving4. Design: A ChatGPT-Style Assistant5. Design: RAG Over Private Documents6. Design: Agents and Tool Use7. AI Gateway and Semantic Caching8. Evaluation and Observability9. Reliability and Cost10. Practice Prompts11. Questions and Answers+1 more - 06Frontend System Design
Frontend and full-stack system design: the RADIO framework, rendering strategies (CSR, SSR, SSG, ISR, streaming, RSC), Core Web Vitals, data fetching and state, real-time and offline, micro-frontends, security, and worked UI designs.
Mid to seniorTable of Contents1. The RADIO Framework2. Rendering Strategies3. Performance and Core Web Vitals4. Data Fetching and State5. Real-Time and Offline6. Component and Application Architecture7. Micro-Frontends8. Security9. Accessibility and Internationalization10. Observability and Release11. Design: News Feed UI+4 more - 07Reliability and Operations
Senior-level system design operations: SLOs and error budgets, observability with OpenTelemetry, deployment strategies, disaster recovery, security architecture, incident response, and lessons from the October 2025 AWS and November 2025 Cloudflare outages.
SeniorTable of Contents1. SLIs, SLOs and Error Budgets2. Observability3. Deployment Strategies4. Backups and Disaster Recovery5. Capacity, Load Testing and Cost6. Security Architecture7. Incident Response8. Case Study: AWS US-EAST-1, October 20259. Case Study: Cloudflare, November 202510. Lessons You Can Reuse11. Production Readiness Checklist+2 more
Read Fundamentals first if you have not. For the class-design side of interviews, see LLD.