Zerops RAG Infrastructure Example

A hello-world demonstration showing how Zerops handles the infrastructure complexity of RAG (Retrieval Augmented Generation) applications — so you can see how to integrate it into your own projects.

What This Is (And Isn't)

This is: A minimal example demonstrating Zerops' capabilities for AI/LLM infrastructure
This isn't: A RAG starter template or production-ready application

We intentionally kept the AI logic dummy-simple (first 500 chars, fake embeddings) to focus on what matters: showing how Zerops eliminates DevOps complexity.

The Infrastructure Challenge

Modern RAG applications need:

• Vector database (Qdrant)
• Relational database (PostgreSQL)
• Caching layer (Valkey/Redis)
• Message queue (NATS)
• Object storage (S3-compatible)
• API service + Workers
• Load balancing, SSL, scaling, monitoring...

Setting this up traditionally requires extensive DevOps knowledge. With Zerops, it's just YAML configuration.

What This Example Shows

1. One-Click Infrastructure

Deploy 8 interconnected services in 60 seconds:

┌─────────────┐     ┌─────────────┐     ┌─────────────┐
│   Qdrant    │     │ PostgreSQL  │     │   Valkey    │
│ (vectors)   │     │ (metadata)  │     │  (cache)    │
└─────────────┘     └─────────────┘     └─────────────┘
       ↑                    ↑                    ↑
       └────────────────────┼────────────────────┘
                           │
┌─────────────┐     ┌─────────────┐     ┌─────────────┐
│  FastAPI    │────▶│    NATS     │────▶│   Worker    │
│   (API)     │     │  (queue)    │     │ (processor) │
└─────────────┘     └─────────────┘     └─────────────┘
       ↑                                         │
       │                                         ▼
┌─────────────┐                         ┌─────────────┐
│  Dashboard  │                         │     S3      │
│   (web)     │                         │  (storage)  │
└─────────────┘                         └─────────────┘

2. Service Auto-Configuration

See how Zerops automatically wires services together using environment variables:

# PostgreSQL connection - no manual config needed
db_pool = await asyncpg.create_pool(
    host=os.getenv("DB_HOST"),  # Zerops provides this
    password=os.getenv("DB_PASSWORD")  # And this
)

# Same for Redis, S3, NATS, Qdrant...

3. Build & Deploy Pipeline

The zerops.yml shows how to:

• Define build and runtime environments
• Configure auto-scaling
• Set up zero-downtime deployments

4. Development Workflow

Local development using cloud services:

# Connect to project's private network
zcli vpn up

# Get all service credentials
zcli env --dotenv > .env

# Your local code now uses Zerops databases

5. Production Patterns

• Async processing with queues
• Caching strategies
• Service discovery
• High availability options

Quick Demo

1. Deploy: Click the button above (60 seconds)
2. Upload: Try uploading a document via the dashboard
3. Search: See how the services interact
4. Explore: Check logs, metrics, and service details in Zerops dashboard

Key Files to Examine

File	What It Demonstrates
zerops-project-import.yml	Development environment setup
zerops-project-production-import.yml	Production HA configuration
zerops.yml	Build & deploy pipeline
api/main.py	Service integration patterns

Understanding the Integration

Project Import Structure

project:
  name: rag-demo
  envVariables:
    # Zerops auto-references service credentials
    DB_HOST: ${db_hostname}
    QDRANT_URL: ${qdrant_connectionString}
    # ... automatic for all services

services:
  - hostname: db
    type: postgresql@16
    mode: HA  # or NON_HA

Build Configuration

# From zerops.yml
- setup: api
  build:
    deployFiles: ./api/~
  run:
    base: python@3.11
    ports:
      - port: 8000
        httpSupport: true

Cost Example

• Development (non-HA): ~$10/month
• Production (full HA): ~$30/month
• Scaling: Pay per minute for actual usage

2-3x cheaper than comparable platforms, with no seat limits or tier restrictions.

The Zerops Advantage

This example demonstrates how Zerops provides:

✅ Complete infrastructure in one YAML file
✅ Automatic service wiring via environment variables
✅ Built-in CI/CD with zero-downtime deploys
✅ Development/production parity with different resource allocations
✅ No DevOps required — focus on your application logic

Learn More

• Full Article: The Perfect Cloud Platform for Development and Production of AI/LLM Apps
• Documentation: Zerops Docs
• YAML Specification: zerops.yml Reference
• Community: Discord

---

This example shows how Zerops handles infrastructure complexity. Use these patterns in your own RAG applications to eliminate DevOps overhead and focus on building your AI features.