Course Overview
Traditional MLOps is a set of practices to productionize traditional ML systems for enterprise applications. Generative AI raises new challenges in managing and productionizing applications at scale. The field of generative AI operations seeks to address these new challenges. In this course, you learn about the challenges that arise when deploying and productionizing generative AI-powered applications. You learn how to secure your generative AI-powered applications. Finally, you will discuss best practices for logging and monitoring your generative AI-powered applications in production.
Who should attend
Developers, DevOps engineers and machine learning engineers who wish to operationalize GenAI-based applications
Prerequisites
Completion of Application Development with LLMs on Google Cloud (ADLGC) or equivalent knowledge.
Course Objectives
- Understand the challenges in productionizing applications using generative AI
- Manage experimentation and evaluation for LLM-powered application
- Productionize LLM-powered applications
- Secure generative AI applications
- Implement logging and monitoring for LLM-powered applications
Outline: Generative AI in Production (GAIP)
Module 1 - Introduction to Generative AI in Production
Topics:
- Generative AI Operations
- Traditional MLOps vs. GenAIOps
- Components of an LLM System
- RAG/ReAct architecture
Objectives:
- Understand generative AI operations
- Compare traditional MLOps and GenAIOps
- Analyze the components of an LLM system
- Define and compare RAG and ReAct
Module 2 - Generative AI Application Deployment
Topics:
- Application deployment options
- Deployment, packaging, and versioning
Objectives:
- Evaluate application deployment options
- Deploy, package, and version apps
Activities:
- Lab: Deploying an Agentic Application on Cloud Run
Module 3 - Productionizing Generative AI
Topics:
- Maintenance and updates
- Testing and evaluation
- CI/CD pipelines for gen AI-powered apps
Objectives:
- Maintain and update LLM models
- Test and evaluate gen AI-powered apps
- Deploy CI/CD pipelines for gen AI-powered apps
Activities:
- Lab: Tracking Versions of Generative AI Applications
Module 4 - Securing Generative AI Applications
Topics:
- Security challenges
- Prompt security
- Sensitive Data Protection and DLP API
- Model Armor
Objectives:
- Identify security challenges for gen AI applications
- Understand prompt security issues
- Apply sensitive data protection and DLP API
- Implement Model Armor
Activities:
- Lab: Securing Generative AI-Powered Applications
Module 5 - Observability for Production LLM Systems
Topics:
- Cloud Operations
- Cloud Logging
- Monitoring
- Cloud Trace
- Agent Analytics and AgentOps
- Putting it all together
Objectives:
- Describe the purpose and capabilities of Google Cloud Observability
- Explain the purpose of Cloud Monitoring
- Explain the purpose of Cloud Logging
- Explain the purpose of Cloud Trace
Activities:
- Lab: Logging, Monitoring, and Agent Analytics