Anthropic Models on Amazon Bedrock
Contact us to book this course
On-Site, Virtual
1 day
This course teaches developers and cloud practitioners to deploy Anthropic's Claude models using Amazon Bedrock's managed infrastructure. Participants learn to build production applications that leverage Claude's advanced reasoning capabilities and tool use for multi-step workflows—all while maintaining enterprise security, governance, and cost controls through Bedrock's fully managed service. The course covers RAG implementation with knowledge bases, prompt engineering techniques, guardrails configuration, and architectural patterns for global deployment across AWS regions.
Course Objectives
Through two hands-on labs, participants build a complete application integrating S3, OpenSearch, Lambda, and CloudWatch with Claude via Bedrock APIs. The course includes practical guidance on model selection (Opus, Sonnet, Haiku), monitoring token consumption and latency, and using Claude Code to accelerate Bedrock integration development. By course end, participants deploy a working system and understand how to design resilient, scalable AI applications using Bedrock's cross-region capabilities.
Who Should Attend
Software Developers, Solutions Architects, DevOps Engineers, Machine Learning Engineers, and Technical Managers building AI-powered applications on AWS.
Prerequisites
- Basic familiarity with AWS console and core services (S3, Lambda, IAM)
- Understanding of REST APIs and JSON
- Programming experience in Python or JavaScript (recommended but not required)
- Command line/terminal comfort for Claude Code section
Course outline
- Generative AI and the Claude model family (Opus, Sonnet, Haiku)
- Investigate Anthropic's Claude models through Amazon Bedrock on AWS
- Bedrock console playground with Claude Sonnet — model access and token tracking
- Regional availability for Claude models on Bedrock
- Cost optimization: balancing region selection, latency, model choice, and quotas
- Resilience patterns: multi-region deployment and failover strategies
- Messages API: requests, responses, system prompts, and conversation structure
- Claude-specific prompt engineering techniques: XML tags for structure, chain-of-thought reasoning, and few-shot examples
-
- Demo: Naive vs. structured prompt comparison using XML tags, chain-of-thought, and multi-turn conversations
- Knowledge bases in Bedrock: data-to-embedding conversion and retrieval-augmented generation (RAG)
- Claude-specific RAG optimization: citation generation, retrieval accuracy, and context utilization
-
- Demo: Connect an S3 bucket to a Bedrock knowledge base, configure ingestion, and test RAG queries with Claude
- Selecting and invoking Claude models (Opus, Sonnet, Haiku) via Bedrock
- Text-generation prompts and measure latency/token usage
- Creating a simple knowledge base from provided documents
- Testing RAG queries and verifying citation accuracy
- Claude's tool use (function calling): defining tools, structured outputs, and multi-step workflows
- Tool definition schemas and response parsing
- Real-world use cases: API integration, data processing, and autonomous task completion
- Demo: Build a multi-step agentic workflow where Claude uses tools to query APIs, process data, and generate structured outputs
- Security fundamentals: encryption, IAM policies, data isolation, and the AWS shared responsibility model
- Bedrock Guardrails: content filtering, topic blocking, PII redaction, and custom policies
- Monitoring: CloudWatch metrics for token usage, latency, errors, and model performance
-
- Demo: Configure Bedrock Guardrails and monitor a Claude endpoint with CloudWatch
- Data ingestion (S3/DynamoDB), knowledge base retrieval, Claude invocation, and monitoring
- Integration patterns using: Lambda, ECS/EKS, and API Gateway
- Model selection, region vs. latency, cost optimization, and resilience strategies
-
- Demo: Walk through a complete architecture diagram and deployment workflow
- Claude Code as a terminal-based agentic coding assistant
- Workflow: scaffolding projects, generating Bedrock integration code, writing tests, and debugging
- Using Claude Code to create a Python/Node.js app that invokes Claude via Bedrock
- Discuss benefits: accelerated development, code quality, and best practices for Bedrock integration
- Introduction to how Amazon uses Claude models for enterprise-scale code generation and architecture
- Claude Code (developer-focused, terminal-based) vs. Kiro (enterprise platform)
- Claude's reasoning capabilities for complex software development tasks
- Use case: chatbot with RAG, document analyzer, or data processing pipeline
- Tools: define custom tools and integrate external APIs
- Guardrails: content filtering and PII protection
- Monitoring: CloudWatch dashboards for performance tracking
- Deployment: serverless or containerized deployment
