feat: Add AI inference infrastructure and OSS agent example

[lukebrady]

🚀 Overview

This PR introduces a comprehensive AI inference infrastructure solution with
modular architecture, automated AMI building, and practical OSS agent examples. The
implementation follows Infrastructure as Code best practices and provides an
"executable codebase" approach where all functionality is discoverable through a
unified Makefile interface.

🏗️ Infrastructure Components

Custom AMI Building (Packer)

• Ubuntu 24.04 base optimized for AI workloads
• GPU support with NVIDIA drivers and container toolkit
• Docker integration with systemd service management
• Automated provisioning scripts for consistent deployments

Modular OpenTofu Configuration

• IAM Module: Dedicated roles, policies, and instance profiles for secure EC2
  access
• Inference Module: EC2 instances with vLLM server deployment and container
  lifecycle management
• Security: EBS encryption, security groups, and IAM best practices
• Scalability: Configurable instance types and storage volumes

🤖 Agent Examples

OSS Agent Implementation

• vLLM Integration: OpenAI-compatible API for local model inference
• Function Calling: Wikipedia search tools with structured Pydantic schemas
• Interactive REPL: Rich formatting with markdown rendering
• Comprehensive Documentation: Inline comments explaining agent patterns and
  OSS model usage

🛠️ Automation & Tooling

Makefile Interface (Executable Codebase)

# Infrastructure Operations
make ami-build                    # Build custom Ubuntu 24.04 AI AMI
make tofu-iam-apply              # Deploy IAM resources
make tofu-inference-apply        # Deploy inference infrastructure
make tofu-apply                  # Deploy all infrastructure

# Agent Operations
make agent-oss-run               # Run OSS agent interactively
make agent-oss-install           # Install dependencies
make agent-oss-check             # Validate environment

# Development & Validation
make check-prerequisites         # Verify required tools
make setup                       # Complete project initialization

CI/CD Pipeline

• GitHub Actions: Automated validation of Makefile commands
• Infrastructure Testing: Packer configuration validation and OpenTofu planning
• Quality Assurance: Continuous integration for infrastructure changes

🔧 Key Features

• Modular Architecture: Separate IAM and inference deployments for better security
  and maintainability
• Container Management: systemd integration with Docker for reliable service
  lifecycle
• GPU Acceleration: L4-equivalent instances (g5.2xlarge) with NVIDIA container
  runtime
• Agent Framework: Practical examples of building agents with OSS models
• Documentation: Comprehensive inline documentation and README updates
• Executable Interface: Unified Makefile serving as a discoverable tool catalog

📊 Changes Summary

• 26 files changed: 2,297 insertions, 2 deletions
• New Infrastructure: Complete OpenTofu modules for IAM and inference
• Custom AMI: Packer configuration for GPU-enabled Ubuntu 24.04
• Agent Examples: Full OSS agent implementation with rich documentation
• Automation: 20+ Makefile commands for streamlined operations
• CI/CD: GitHub Actions workflow for continuous validation

🚀 Quick Start

  # Setup and validate environment
  make setup

  # Build custom AMI
  make ami-build

  # Deploy infrastructure
  make tofu-apply

  # Run OSS agent
  make agent-oss-run

🎯 Design Philosophy

This implementation embodies the "executable codebase" principle where:

• All functionality is discoverable through make help
• Both humans and AI agents can interact with the same interface
• Infrastructure and code are equally documented and maintainable
• Every command is self-contained and includes necessary context

🔄 Future Enhancements

• Configurable hostname and vLLM arguments in user_data template
• Auto-scaling groups for production workloads
• Monitoring and observability stack integration
• Additional agent examples and patterns