https://github.com/sunilgentyala/archforge
The Modern System Design Workbench: 60+ interactive patterns, real-time capacity calculator, AI/LLM architectures, and interview-ready templates.
https://github.com/sunilgentyala/archforge
ai-architecture architecture backend capacity-planning cheatsheet cloud-native design-patterns distributed-systems interview-prep javascript learning llm microservices open-source rag scalability software-engineering system-design system-design-interview vector-database
Last synced: 12 days ago
JSON representation
The Modern System Design Workbench: 60+ interactive patterns, real-time capacity calculator, AI/LLM architectures, and interview-ready templates.
- Host: GitHub
- URL: https://github.com/sunilgentyala/archforge
- Owner: sunilgentyala
- License: mit
- Created: 2026-06-28T20:58:16.000Z (27 days ago)
- Default Branch: main
- Last Pushed: 2026-06-28T21:14:28.000Z (27 days ago)
- Last Synced: 2026-06-28T23:06:15.340Z (27 days ago)
- Topics: ai-architecture, architecture, backend, capacity-planning, cheatsheet, cloud-native, design-patterns, distributed-systems, interview-prep, javascript, learning, llm, microservices, open-source, rag, scalability, software-engineering, system-design, system-design-interview, vector-database
- Homepage: https://sunilgentyala.github.io/archforge/
- Size: 64.5 KB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- Contributing: CONTRIBUTING.md
- License: LICENSE
Awesome Lists containing this project
README
ArchForge
The Modern System Design Workbench
Live Site
·
60+ Patterns
·
Capacity Planner
·
Templates
·
Contribute
---
## What is ArchForge?
Most system design resources fall into one of two buckets: a static wall of markdown that you read once and forget, or a paywalled course that recycles the same eight example problems. The most popular free reference covers roughly 20 patterns, was last meaningfully updated around 2017, and predates the entire AI/ML infrastructure space.
ArchForge was built to fill those gaps. It is a browser-based workbench with no install, no account, and no cost:
- **60+ interactive pattern cards**, each with a trade-off breakdown and complexity rating
- **A live capacity planner** that calculates RPS, storage, bandwidth, and cache size from sliders
- **7 AI and LLM architecture patterns** covering things like RAG pipelines, LLM serving, multi-agent orchestration, and model routing
- **6 fully worked system design templates** for interviews and production planning
- **Modern patterns** that simply did not exist when the primer was written: vector databases, service meshes, cell-based architecture, edge computing, serverless
Everything is open source and will stay that way.
---
## Live Site
> **[sunilgentyala.github.io/archforge](https://sunilgentyala.github.io/archforge/)**
>
> No signup. No paywall. No ads. Works in any browser.
[](https://sunilgentyala.github.io/archforge/)
---
## Where ArchForge Fills the Gaps
| Feature | ArchForge | System Design Primer | ByteByteGo | Grokking |
|---|:---:|:---:|:---:|:---:|
| AI / LLM Architecture Patterns | 7 patterns | None | Partial | None |
| Interactive Capacity Calculator | Built-in | Static tables | No | No |
| Total Patterns Covered | 60+ | ~20 | Many | ~15 |
| Edge / Serverless Patterns | Yes | No | Partial | No |
| Vector Database Pattern | Yes | No | Partial | No |
| Architecture Templates | 6 detailed | 8 solutions | Yes | Yes |
| Free and Open Source | Yes | Yes | Freemium | Paid |
---
## Pattern Library
60+ patterns across 8 categories. Each card shows the tagline and complexity rating. Click any card to expand the full detail: what it is, when to reach for it, what you gain, and what you give up.
| Category | Count | Patterns |
|---|:---:|---|
| Infrastructure | 10 | Load Balancer, CDN, API Gateway, Service Mesh, Reverse Proxy, DNS Load Balancing, Edge Computing, Container Orchestration, Serverless Functions, Service Discovery |
| Data Storage | 12 | Master-Replica Replication, Multi-Master Replication, Horizontal Sharding, Database Federation, Connection Pooling, Time-Series Storage, Graph Database, Vector Database, Data Lake, Data Warehouse, Write-Ahead Log, MVCC |
| Caching | 8 | Cache-Aside, Write-Through, Write-Behind, Read-Through, Cache Invalidation Strategies, Distributed Cache, In-Process Cache, CDN Cache Headers |
| Scalability | 8 | Horizontal Scaling, Vertical Scaling, Auto-Scaling Policies, Queue-Based Load Leveling, Back-Pressure, Rate Limiting, Bulkhead Isolation, Cell-Based Architecture |
| Reliability | 6 | Circuit Breaker, Retry with Exponential Backoff, Saga Pattern, Idempotency Design, Blue-Green Deployment, Canary Release |
| API and Messaging | 8 | REST, GraphQL, gRPC, WebSocket, Server-Sent Events, Message Queue, Event Streaming, Webhook Design |
| AI / ML Systems | 7 | LLM Serving Architecture, RAG Pipeline, Feature Store, ML Training Pipeline, Embedding Search, Multi-Agent Orchestration, Model A/B Testing |
| Security | 5 | Zero Trust Architecture, OAuth 2.0 + PKCE, JWT Authentication, Secrets Management, API Rate Limiting |
---
## Capacity Planner
Adjust five sliders and get six live estimates instantly. No spreadsheet, no mental math.
**Inputs**
- Daily active users
- Average requests per user per day
- Average request and response size
- Storage per user per day
- Data retention period in years
**Outputs**
- Average RPS and 3x peak RPS
- Daily storage growth
- Total storage over the retention window
- Peak bandwidth
- Cache size based on the 10% hot data heuristic
---
## Architecture Templates
Six complete system designs, each with a component list, scale targets, data model sketch, and the specific trade-offs worth mentioning out loud in an interview.
| Template | Difficulty | Scale Target |
|---|:---:|---|
| URL Shortener | Beginner | 10B URLs, 100K redirects/s |
| Social Media Feed | Intermediate | 500M DAU, push/pull hybrid fan-out |
| Video Streaming | Advanced | 2B DAU, adaptive bitrate HLS |
| Chat System | Intermediate | 2B DAU, 50B messages/day |
| Ride Sharing | Advanced | 10M live trips, 4Hz driver location |
| LLM API Service | Advanced | 10M requests/day, prompt caching, model tiering |
Template markdown files live in [`/templates`](templates/) and can be read directly without opening the browser app.
---
## Running Locally
There is no build step. The entire app is a single HTML file.
```bash
# Clone the repo (useful if you found this via the live site or a link elsewhere)
git clone https://github.com/sunilgentyala/archforge.git
cd archforge
```
Then open the app in one of two ways:
```bash
# Option 1 -- open the file directly in your default browser
# macOS
open docs/index.html
# Windows
start docs/index.html
# Linux
xdg-open docs/index.html
# Option 2 -- serve locally (recommended, ensures relative paths work correctly)
python -m http.server 8080 --directory docs
# Then visit http://localhost:8080
```
---
## Repository Layout
```
archforge/
├── docs/
│ ├── index.html # The full interactive app (one file, no dependencies)
│ ├── assets/
│ │ └── logo.svg # SVG brand mark
│ └── og-image.svg # Social sharing thumbnail (1200x630)
├── patterns/
│ └── README.md # Pattern authoring guide and schema reference
├── templates/
│ ├── url-shortener.md
│ ├── social-media-feed.md
│ ├── video-streaming.md
│ ├── chat-system.md
│ ├── ride-sharing.md
│ └── llm-api-service.md
├── CONTRIBUTING.md
└── LICENSE
```
---
## Contributing
Adding a pattern takes roughly 15 minutes. You write one JavaScript object, verify it looks right in a browser, and open a PR. No build tools, no npm install.
See [CONTRIBUTING.md](CONTRIBUTING.md) for the full pattern schema, complexity scale, writing style guide, and review checklist.
Areas where contributions are especially welcome:
- Cloud provider-specific patterns (AWS, Azure, GCP)
- FinTech and HealthTech system designs
- Database engine internals
- Additional AI and ML production patterns
---
## License
MIT. Use it, fork it, teach with it, ship with it.
---