Building Generative AI Applications
summary
Shows how to attach large language models to real data using retrieval-augmented generation with Haystack, combining an embedding retriever + prompt templates and a Haystack prompt node to inject retrieved documents so answers stay factual. Uses Amazon Bedrock, SageMaker JumpStart, OpenSearch (or Pinecone), SQS batching, Kubernetes + KEDA for throughput, NVIDIA device plugin for GPU scheduling, and mentions TRN1 giving ~50% faster training and Inferentia ~40% faster inference. Targets millions of files and thousands of pipelines.
AI-written summary · about
chapters (7)
- 0:00Introduction and welcome
- 1:09What is generative AI on AWS
- 5:48Generative AI ecosystem overview
- 9:44Fine-tuning generative models
- 13:56Building apps with Haystack
- 25:19NLP inference at scale
- 34:02Q&A and resources
related talks
28mImproving responses in generative AI applications
29mComposing AI Applications as a Graph with Haystack
53mGenerative AI: Overview and use cases
15mHarnessing the Power of Generative AI for Intelligent Applications
44mDecoding generative AI
07mBuild Generative AI Apps for Enterprises with Cloud
29mCustomizing genAI applications for your business
46mGenerative AI and Java