System Design 101 — From Level 0 to Advanced

A complete, self-paced curriculum in system design, written as a durable archive. Every module is a standalone deep-dive with theory, worked numbers, diagrams, a quiz, and a Q&A section.

Audience: engineers who can code but have never been taught distributed systems formally. Starts at absolute zero (what is a server?) and ends at designing planet-scale and AI/ML inference systems.

Method: classical (non-AI) examples are used to teach fundamentals, because they are the cleanest teaching vehicles. Where a concept maps onto ML/LLM infrastructure, a 🤖 AI/ML Callout section makes the connection explicit. Part VII is dedicated entirely to AI systems.

---

How to use this repo

If you want to...	Do this
Learn from scratch	Read modules in order. Do not skip Part I.
Revise before an interview	Read the TL;DR + Cheat Sheet at the top/bottom of each module
Self-test	Each module ends with a quiz; answers are in <details> blocks
Look up a term	See glossary.md
Estimate capacity	See cheatsheets/numbers.md
Read real-world engineering write-ups alongside a module	See cheatsheets/further-reading.md

Progress tracker: PROGRESS.md

---

Curriculum

Part 0 — Foundations of Thinking

#	Module	Status
00	Mental Models, Vocabulary & Back-of-the-Envelope Math	✅

Part I — The Machine and The Network

#	Module	Status
01	Inside One Machine: CPU, Memory, Disk, OS, Concurrency	✅
02	Networking for System Designers: IP, TCP/UDP, DNS, TLS, HTTP/1-2-3	✅
03	API Design & Communication Styles: REST, RPC, gRPC, GraphQL, WebSockets	✅

Part II — Storage

#	Module	Status
04	Storage Engine Internals: B-Trees, LSM-Trees, WAL, Indexes, Erasure Coding	✅
05	Relational Databases & Data Modeling	✅
06	Transactions: ACID, Isolation Levels, MVCC, Pessimistic/Optimistic Locking	✅
07	The NoSQL Landscape: KV, Document, Wide-Column, Graph, Time-Series, Vector	✅
08	Caching: Patterns, Eviction, Redis, CDNs, Stampedes	✅

Part III — Distribution

#	Module	Status
09	Replication: Leader/Follower, Multi-Leader, Leaderless, Quorums	✅
10	Partitioning & Sharding: Consistent Hashing, Rebalancing, Hotspots	✅
11	CAP, PACELC & Consistency Models	✅
12	Consensus, Coordination & Time: Raft, etcd, Distributed Locks, Failure Detection (Heartbeats/Gossip/Phi-Accrual), Vector Clocks	✅
13	Distributed Transactions: 2PC, Sagas, Outbox, Idempotency	✅

Part IV — Scale and Traffic

#	Module	Status
14	Load Balancing, Proxies & Service Discovery	✅
15	Messaging & Streaming: Queues vs Logs, Kafka, Delivery Semantics	✅
16	Batch & Stream Processing: MapReduce, Spark, Flink, CDC, Event Sourcing	✅
17	Rate Limiting, Quotas & Admission Control	✅
18	Resiliency: Timeouts, Retries, Circuit Breakers, Degradation	✅

Part V — Running It in Production

#	Module	Status
19	Observability: Metrics, Logs, Traces, SLI/SLO/Error Budgets	✅
20	Deployment & Infrastructure: Containers, K8s, Canary, Multi-Region, DR	✅
21	Security: AuthN/AuthZ, OAuth/OIDC, Encoding vs Encryption vs Tokenization, Password Storage, Multi-Tenancy	⬜

Part VI — Architecture at Large

#	Module	Status
22	Monoliths, Microservices, DDD, Gateways & Service Mesh	⬜
23	Data Architecture: OLTP vs OLAP, Warehouses, Lakes, Lakehouses	⬜

Part VII — AI/ML Systems

#	Module	Status
24	ML Systems Foundations: Training/Serving Split, Feature Stores, Skew	⬜
25	Inference at Scale: Batching, KV Cache, GPU Scheduling, Autoscaling	⬜
26	Vector Search & RAG Systems: HNSW/IVF, Hybrid Search, Freshness	⬜
27	LLM Application Architecture: Agents, Guardrails, Evals, Cost Control	⬜

Part VIII — Interview Craft & Case Studies

#	Module	Status
28	The Interview Framework: A Repeatable 45-Minute Structure	⬜
29	Case Study: URL Shortener + Distributed ID Generation	⬜
30	Case Study: Chat System (WhatsApp/Slack)	⬜
31	Case Study: News Feed / Timeline (Twitter, Instagram)	⬜
32	Case Study: Video Platform (YouTube/Netflix)	⬜
33	Case Study: Geospatial & Ride Hailing (Uber)	⬜
34	Case Study: Search, Typeahead & Web Crawler	⬜
35	Case Study: Payments & Double-Entry Ledger	⬜
36	Case Study: Object Store (S3) & Distributed File Systems	⬜
37	Case Study: Collaborative Editing (OT vs CRDT)	⬜
38	Case Study: Ad Click Aggregation & Real-Time Analytics	⬜
39	Case Study: Recommendation System at Scale	⬜
40	Case Study: LLM Serving Platform (ChatGPT-scale)	⬜
41	Case Study: Enterprise RAG over 100M Documents	⬜
42	Bonus Case Study: Proximity Service & Geospatial Indexing (Quadtree/Geohash)	⬜
43	Bonus Case Study: Design Gmail (Email at Scale)	⬜

Part IX — Curated Industry Case Studies (Supplementary Reading)

Not standalone modules — these are real engineering blog write-ups mapped to the module they reinforce, to read after that module so the theory has somewhere concrete to land. Tracked in cheatsheets/further-reading.md.

---

Conventions used in these notes

• ⚠️ Trap — a common misconception or interview failure mode.
• 🤖 AI/ML Callout — how the concept maps to ML/LLM infrastructure.
• 📐 Numbers — quantitative reasoning you should be able to reproduce.
• 🎤 Interview Angle — how this is probed in an interview.

---

Maintained as a personal learning archive. Feel free to fork.

---