Learn how Uber, Netflix, Stripe, and modern AI agents are built, one system at a time.
A practical, diagram-first guide for engineers who build real systems and ace system design interviews.
Start here · What's new · Case studies · Theories · Interview prep · 简体中文 · Website
⭐ Star this repo to follow new chapters as they land.
The 2nd Edition is a work in progress. New chapters are published regularly, so watch the repo for updates.
Since 2015, System Design and Architecture has helped software engineers around the world land roles and design better systems. It works because it is:
- 🌎 Grounded in real systems. Every design starts from how companies such as Uber, Facebook, Netflix, Stripe, and Dropbox actually build their products.
- 🍕 Easy to digest. System design at the right level of abstraction is a map that gets you to the destination by the shortest route. Charts, diagrams, and step-by-step guides give you the minimum actionable knowledge to excel in interviews and start building.
- 🤲 Broad. It covers backend to frontend, and classic distributed systems to LLM inference and AI agents.
- 🚶♀️ Repeatable. Every design follows the same 4-step framework:
- Clarify requirements
- Sketch the high-level design
- Deep dive into components and how they interact
- Wrap up with blind spots and bottlenecks
| If you are... | Read this path |
|---|---|
| 🧑💼 Senior (Google L5+) | Go straight to System Design in Practice. |
| 🧑💻 Junior or mid-level (Google L3/L4) | Read System Design Theories first, then System Design in Practice. |
| 🌱 Just starting out | Begin with Prepare for an Interview effectively, then Theories, then Practice. |
The 2nd Edition covers what changed between 2015 and today: mobile ate the world, cloud-native computing arrived with Kubernetes, web3 rose and fell, and AI agents became a new class of system. Recent additions:
Also coming in the 2nd Edition: more companies and products, existing chapters rewritten for current trends, more diagrams with step-by-step guides, a PDF edition, and less repetition so that each design teaches its own abstraction.
Out of scope: API design for specific domains (see Google's API design guide) and deep object-oriented design.
| Product | Question |
|---|---|
 |
Designing an Always-on Agent |
 |
Designing a recommendation system |
 |
Fraud Detection with Semi-supervised Learning |
Foundations
- Introduction to Architecture
- How to scale a web service?
- Concurrency Model
- Cloud Design Patterns
- Public API Choices
- Load Balancer Types
- How to do capacity planning? (coming soon)
Data and consistency
- ACID vs BASE
- Data Partition and Routing
- Replica, Consistency, and CAP theorem
- Improving availability with failover
- Intro to Relational Database
- 4 Kinds of No-SQL
- Key value cache
- Stream and Batch Processing Frameworks
- Lambda Architecture
Data structures
AI systems
Engineering craft
- Crack the system design interview
- What can we communicate in soft skills interview?
- Experience Deep Dive
- Introduction to software engineer interview (coming soon)
- How to crack the coding interview, for real? (coming soon)
- Culture fit (coming soon)
- Be a software engineer - a hero's journey (coming soon)
- 🇺🇸 English: you are here
- 🇨🇳 简体中文
Tian Pan is a software engineer and engineering manager who has worked in the San Francisco Bay Area for 10+ years, at Uber, Oracle, IoTeX, and Microsoft. More notes live at tianpan.co.
- 💬 Join the discussion on Telegram or Discord.
- 🐛 Found a typo, a broken diagram, or a gap in a design? Open an issue or send a PR.
- 🌐 Translations are welcome. Chapters in
en/andzh-CN/use identical filenames.
GPL v3
Found this useful? ⭐ Star the repo. Spotted a gap? Open a PR.