Scalable Kubernetes Infrastructure for AI Platforms

Chapter 6. Summary and Outlook

In this report, you learned about the multilayered lifecycles that govern AI projects (the AI development lifecycle in Figure 1-1 and the AI model lifecycle in Figure 3-1) and open source Kubernetes-based tools that work to enable each of those phases at scale. You learned how to leverage open source tools to successfully move a generative AI model through these cycles, standardizing the process of model creation and allowing you to confidently deploy and manage AI models in production.

To understand how these phases interact, which open source tools to use at each phase, and how they look in practice, let’s look at an example project.

Personalized Healthcare Chatbot

In this example, let’s follow a fictional generative AI team at a major health insurer as it pitches and builds a personalized health chatbot. In the first phase of the AI development lifecycle, the project is initiated.

During project initiation, the generative AI team lead meets with the team’s organization’s director of engineering, head of sales, head of IT, and director of research and development to discuss a project idea floated by a member of the team. She pitches a personalized chatbot that healthcare subscribers can interact with to navigate questions about their personal health, their insurance policy, and healthcare providers. She claims this can reduce time spent by customer agents on common, personalized tasks; reduce personally identifiable information (PII) and health ...