Deploying Retrieval-Augmented Generation Applications with Ray, Hugging Face, and LangChain on Google Cloud

Learn how to deploy a fully-functional Retrieval-Augmented Generation (RAG) application to Google Cloud using open-source tools and models from Ray, HuggingFace, and LangChain in this 41-minute conference talk from Google Cloud Next 2024. Discover techniques for augmenting the application with custom data using Ray on Google Kubernetes Engine (GKE) and Cloud SQL's pgvector extension. Explore the process of deploying any model from HuggingFace to GKE and rapidly developing LangChain applications on Cloud Run. Gain insights into core considerations, technology learnings, implementation strategies, scaling characteristics, serving subsystems, security measures, and data sensitivity. Speakers Alex Zakonov, Brandon Royal, and Stephen Allen cover topics including problems with large language models, the benefits of Kubernetes, G Appliances, customer feedback, natural language APIs, Identity Wear Proxy, and GK Quick Start. By the end of the session, acquire the knowledge to deploy and customize your own RAG application to meet specific needs.

Introduction
Problems with large language models
Core things to consider
Why kubernetes
G Appliances
Customer feedback
Technology learnings
Technology
Implementation
Scaling characteristics
Serving subsystem
LangChain
Security
Data sensitivity
Natural language API
Identity Wear Proxy
GK Quick Start
Jupiter Hub