GenAI Bootcamp: Production

Getting Started

Please follow the getting started guide to get your machine configured.

Opening code

Because this is a monorepo, VS Code/Cursor and the like can be slow to load, because it needs to run checks on the Python code for every single project.

Instead, load only the code for the task you're working on by navigating to File -> Open Workspace from File... and selecting the workspace file (eg week5/1-hello-world/1-hello-world.code-workspace).

---

Configuring Github Actions

You can deploy projects from GitHub actions. You must run from your own fork to do so.

Set up

1. Deploy the CloudFormation stack

   This command will create the IAM role that GitHub Actions will use to deploy your stacks. Replace <your-org> and <your-repo> with your GitHub organization and repository name.

aws cloudformation deploy \
    --template-file github-actions.yaml \
    --stack-name github-actions-role \
    --parameter-overrides GitHubOrg=<your-org> GitHubRepo=<your-repo> \
    --capabilities CAPABILITY_NAMED_IAM

After deployment, get the RoleArn output from the stack.

aws cloudformation describe-stacks \
    --stack-name github-actions-role \
    --query "Stacks[0].Outputs[?OutputKey=='RoleArn'].OutputValue" \
    --output text

2. Set the AWS_ROLE_ARN in GitHub

The RoleArn from the previous step needs to be configured as a repository variable in GitHub.

• Navigate to your repository on GitHub.
• Go to Settings > Secrets and variables > Actions.
• Click New repository secret.
• Name the variable AWS_ROLE_ARN.
• Paste the RoleArn value from the previous step.
• Click Add variable.

Run a deployment from GitHub Actions

• Navigate to the Actions tab in your GitHub repository.
• In the left sidebar, select the workflow you want to run (e.g., Deploy Hello World).
• Click the Run workflow dropdown, and then click the Run workflow button to start the deployment.

---

Topics

Week 5

• Automated deployments and CI/CD with agentic workflows
• Building agentic applications with AWS CDK
• Creating and using tools in agentic applications
• Integrating MCP

Week 6

• GenAI Security
• GenAI security techniques
• Creating and using AI Guardrails
• Beyond chat: Automated document processing workflows

Week 7

• Automated document processing and embedding creation
• Effective RAG use

Week 8

• Capstone project: Building a real production application