aws-samples

End to end example of a Retail Agent implemented with agents for Amazon Bedrock

Curated archive of agentic AI samples, reference implementations, and tooling from the Agentic AI Acceleration program in Korea. AWS-native patterns covering Bedrock, AgentCore, and end-to-end agentic AI pipelines.

Learn EKS Auto Mode through deployable Terraform examples

AIDLC × AgenticOps — plugin marketplace that extends Claude Code and Kiro with agentic operations for the AWS AI-Driven Development Lifecycle (Inception → Construction → Operations).

Serverless event-driven architecture enabling engineering teams to process millions of daily events with near real-time visibility and strong resilience.

Conversational AWS cost estimator. Sample reference architecture using Amazon Bedrock + Claude with the AWS Pricing API and Savings Plans API for live, tool-using cost estimates. Frontend on ECS Fargate, CloudFront + WAF, optional Cognito auth, GitOps CodePipeline.

Sample Amazon Bedrock proxy demonstrating how developers can use any AI framework (LangChain, LangGraph, Strands Agents, CrewAI) through a centrally governed API Gateway with Cognito JWT authentication, request tracking, and response streaming — all transparent to the framework via boto3's event system.

KRO Deploy complete serverless applications on AWS with a single kubectl apply — using Kro, ACK, EKS Capabilities, and Crossplane on Kubernetes (Hybrid Kro Compositions pattern)

Sketch-to-3D is an end-to-end Gen AI pipeline deployed on AWS transforming 2D sketches into photorealistic product images and interactive 3D models.

Multi-agent AI system for autonomous retail pricing optimization using Amazon Bedrock AgentCore. 6 specialized agents orchestrate competitive intelligence, demand forecasting, and strategy synthesis to generate ranked pricing scenarios with human-in-the-loop governance.

Sample multi-agent AI system for life insurance claims processing using Amazon Bedrock AgentCore. Six specialist agents orchestrate identity verification, document extraction, policy validation, fraud detection, and adjudication with human-in-the-loop governance.

Sample code for two-tier LLM inference (cascading) on Amazon Bedrock, using a single Bedrock Converse API surface. NVIDIA Nemotron Nano handles routine support-ticket classification on every request; Anthropic Claude Sonnet handles cases the routing logic flags as harder. Includes a bake-off harness, sample data, and tests. Next.js + TypeScript.

Reference architectures for digital asset tokenization on AWS. Sample code for financial messaging integration, settlement, price feeds, proof-of-reserves, and stablecoin payments.

Sample multi-service insurance claims processing app on AWS — ECS Fargate, Aurora PostgreSQL, and Amazon Bedrock — used to demonstrate autonomous incident detection and investigation with AWS DevOps Agent.

Monitor LLM predictive uncertainty in production using conformal prediction on AWS — calibration, real-time CloudWatch monitoring, human-in-the-loop routing, batch evaluation. Built on SageMaker AI, Lambda, DynamoDB, SQS, SNS.

Intentionally vulnerable sandbox app for demonstrating AWS agent-based code review, penetration testing, and 5xx incident investigation on ECS Fargate, ALB, DynamoDB, and CloudWatch. Educational, non-production, MIT-0.

Reduce manual compliance review from days to minutes — orchestrate specialist AI agents to assess documents against regulatory frameworks, standards, or organisational policies. Built on Amazon Bedrock and AgentCore with traceable findings and iterative coaching.

A deployable AWS CDK (TypeScript) sample that stands up an AWS Network Firewall environment (logging, CloudWatch alarms, a least-privilege agent role, and an auto-created DevOps Agent space) with three reproducible break-fix scenarios that AWS DevOps Agent investigates using native AWS APIs.