Generative AI - Architectures and applications in depth
summary
Walks through a concrete RAG pipeline built on AWS Bedrock: ingest PDFs to S3, create embeddings with the Amazon embedding model, store vectors in OpenSearch Serverless, and run a Bedrock agent that calls Lambda-backed action APIs to create an investment portfolio. Demo uses Anthropic LLMs for generation, shows source citations and an execution trace for each action, and reports that RAG avoids costly fine‑tuning while producing accurate, auditable answers for the sample financial dataset.
AI-written summary · about
chapters (15)
- 0:00Generative AI intro
- 1:10ML and deep learning
- 2:35Foundation models importance
- 4:42Large language models and Transformers explained
- 7:31Customizing foundation models
- 9:16Prompt engineering vs fine-tuning
- 10:42Retriever augmented generation
- 12:27RAG workflow and embeddings
- 16:10Amazon Bedrock
- 18:20Setting up knowledge base
- 21:56AI agents intro
- 23:24AI agent components
- 27:35Investment portfolio agent demo
- 34:41Testing agent workflow
- 37:11Summary and next steps
related talks
28mImproving responses in generative AI applications
44mEnhance Your Generative AI Applications with RAG
24mBuilding GenAI applications with Amazon Bedrock
19mMind your business: Secure your generative AI application on AWS
20mExplore what’s possible with new AWS generative AI services
41mServerless LLM with AWS Lambda and LangChain: A Revolution in Application Development
48mHow to Make Smart Architecture Decisions when Building Gen AI Apps
31mLamRAG: From Data to Constructive Insights using Amazon Bedrock