# 05 - AWS Cloud Infrastructure — Home
**Purpose:** AWS architecture, networking, and cost optimization for AI workloads.

## Overview

This section covers our AWS cloud infrastructure setup for scalable AI services, including:
- VPC networking architecture
- EC2 instance optimization for GPU workloads
- S3 data pipeline automation
- Cost optimization strategies

## Infrastructure Components

### **AWS Architecture**
- **US-East-1 (N. Virginia):** Primary production environment
- **Multi-AZ deployment:** High availability for critical services
- **VPC networking:** Isolated subnets for compute, storage, and monitoring

### **Compute Resources**
- **p4d.24xlarge:** A100-based instances for training
- **g5.xlarge:** A10 instances for inference
- **m5.large:** CPU-based instances for web services
- **Spot instances:** Cost savings for non-critical workloads

### **Storage**
- **S3 buckets:** Model weights, datasets, backup archives
- **EBS volumes:** High-performance storage for active VMs
- **Glacier:** Cold storage for historical data

## Cost Optimization

- **Reserved Instances:** 1-3 year commitments for baseline workloads
- **Spot Instances:** 70% savings for fault-tolerant tasks
- **S3 Lifecycle Policies:** Automatic tiering to Glacier
- **CloudWatch Alarms:** Budget alerts and anomaly detection

## Related Documents

- **[Qwen3.5-MoE Setup](../02\ -\ TECHNICAL\ INFRASTRUCTURE/04\ -\ Qwen3.5\ Model\ Setup/)** — Model deployment strategies

---

*Our AWS infrastructure supports our SMB AI services and agent deployments with maximum efficiency.*