RAG and Fine Tuning With Kubeflow

Learn to implement production-ready Retrieval Augmented Generation (RAG) applications using Kubeflow in this 30-minute conference talk from CNCF. Discover how to leverage Kubeflow Trainer, KServe, and Feast to fine-tune, deploy, and serve scalable RAG systems through practical demonstrations. Explore the foundational 2020 RAG paper and understand how Kubernetes and distributed computing enable fine-tuning of both Retrieval and Generator models for knowledge-intensive tasks. Master the deployment process using Feast for feature serving and KServe for model serving. Address critical production AI challenges including data preprocessing and transformation across varying formats, data ingestion pipelines, data lineage tracking, write pattern trade-offs, horizontal scaling strategies, domain-specific model customization, and serving latency optimization. Gain hands-on insights into building end-to-end RAG solutions that can scale in production environments using cloud-native technologies.