System Design Cheat Sheet: 20 Mermaid Diagram Templates for Common Questions

Every system-design interview asks for the same three artifacts: a high-level architecture diagram, a request-path sequence, and a data model. What changes is the problem statement wrapped around them. Below are 20 Mermaid templates for the most common problems — copy the diagram, edit the labels, and you have your starting point.

If you're running Interview Helpers the AI returns Mermaid inline as SVG so you can literally screenshot the diagram to the interviewer's whiteboard. Everything below is the same syntax.

Diagrams 1–10 (with copy-paste Mermaid)

flowchart LR
  U[User] -->|POST tweet| G[API Gateway]
  G --> T[Tweet Service]
  T --> DB[(Tweet DB)]
  T --> Q[Fan-out Queue]
  Q --> F[Fan-out Worker]
  F --> RC[(Timeline Cache)]
  RC --> R[Read Service]
  R --> U

flowchart LR
  U[User] --> R[Redirect Service]
  R -->|hot| C[(Cache)]
  R -->|miss| DB[(URL DB)]
  U --> W[Write Service]
  W --> B[Base62 Generator]
  W --> DB
  W --> A[Analytics Queue]

flowchart LR
  U[Viewer] --> CDN[CDN Edge]
  CDN -->|miss| O[Origin]
  U --> M[Metadata Service]
  M --> DB[(Video DB)]
  U --> H[History Service]
  H --> HDB[(History DB)]
  UP[Uploader] --> T[Transcode Pipeline]
  T --> S[(Object Store)]
  S --> O

flowchart LR
  A[User A] -->|WS| G[Gateway]
  G --> M[Message Service]
  M --> Q[Message Queue]
  Q --> D[Delivery Worker]
  D -->|WS| B[User B]
  M --> DB[(Message Store)]
  A --> P[Presence Service]
  B --> P

flowchart LR
  R[Rider] --> G[API]
  G --> DS[Dispatch Service]
  DS --> GEO[(Geo Index)]
  DS --> M[Matching Engine]
  M -->|assign| D[Driver App]
  D --> GEO
  DS --> E[ETA Service]
  E --> R

flowchart LR
  U[Client] --> G[API Gateway]
  G --> L[Rate Limiter]
  L -->|INCR key| R[(Redis)]
  L -->|allow| S[Service]
  L -->|deny| U

flowchart LR
  U[Client] --> CH[Consistent Hash]
  CH --> N1[(Node A)]
  CH --> N2[(Node B)]
  CH --> N3[(Node C)]
  N1 <-.replica.-> N2
  U -->|miss| DB[(Primary DB)]

flowchart LR
  U[User] --> AC[Autocomplete Service]
  AC --> T[(Trie In-Memory)]
  AC --> RK[(Ranking Store)]
  UP[Upstream Signals] --> K[Kafka]
  K --> W[Update Worker]
  W --> T
  W --> RK

flowchart LR
  S[Producer] --> G[Notify API]
  G --> Q[Notification Queue]
  Q --> D[Dispatcher]
  D --> PN[Push Worker]
  D --> EM[Email Worker]
  D --> SM[SMS Worker]
  PN --> DDB[(Dedupe Store)]

flowchart LR
  M[Merchant] --> G[Payment API]
  G -->|idempotency-key| IDEM[(Idempotency Store)]
  G --> L[Ledger]
  G --> A[Acquirer Adapter]
  A --> B[Bank Network]
  A --> W[Webhook Delivery]
  W --> M

Diagrams 11–20 (concept notes)

These follow the same pattern — feel free to sketch them out in Mermaid the same way.

The 4-diagram interview structure

1. Flowchart — the components and edges. Do this first.
2. Sequence diagram — the read path and the write path. Talk through eventual consistency here.
3. ER model — tables and relationships. Discuss sharding key.
4. Capacity envelope — QPS, storage per year, hot key risk. One paragraph is enough.

The interviewer's grade sheet has one row per artifact. Get all four out in the first 20 minutes and you're passing before you start optimizing.

Ready to have the AI render these directly in the overlay? Grab Interview Helpers and flip on System Design mode. Also see the deeper walkthrough at /for-system-design.