Скачать или смотреть Integrating Databricks AI Gateway with Kong and any external AI Gateway

Lara, Maria and Hari discuss the integration of Databricks AI Gateway with Kong, exploring the architecture, API management, and the process of making API calls to OpenAI and Databricks. They delve into the functionalities of Kong, including traffic management and monitoring, and conclude with insights on future directions for model serving.

Databricks can create model serving endpoints for various platforms.
Mosaic AI Gateway is used for managing Databricks endpoints.
Kong serves as an external gateway for integrating AI models.
API management in Kong includes creating routes and plugins.
Authentication can be managed through service principals and tokens.
Traffic management is crucial for monitoring API calls.
Kong allows for the addition of various plugins for enhanced functionality.
Databricks endpoints can be linked with external gateways like Kong.
The integration supports both Databricks and external model serving.
Future features may include custom external endpoints for greater flexibility.
titles

Chapters
00:00 Introduction to Databricks AI Gateway and Kong Integration
01:49 Understanding the Problem Statement and Objectives
05:06 Architecture Overview of Kong and Databricks Integration
08:22 Demonstrating Kong's AI Proxy Plugin
13:22 Integrating Databricks Model Serving with Kong
17:02 Exploring the Flexibility of Model Hosting

#Databricks #AIGateway #Kong #ModelServing #Integration #APIManagement #OpenAI #Architecture #TrafficManagement #DeveloperPortal