AWS, Machine Learning & Amazon Bedrock Engineering

• AWS Global Infrastructure & Cloud Architecture Principles
  

  Topics
  • AWS global regions, availability zones, edge locations
  • Designing for high availability and fault tolerance
  • Well-Architected Framework pillars
  • Multi-account strategy and landing zones

• Subtopics
  • Shared responsibility model
  • Identity boundaries and service control policies
  • AI workload placement strategy
• Hands-on / Lab
  • Configure multi-account setup using AWS Organizations
  • Implement IAM role-based access with least privilege
• Real-world application
  • Designing a secure enterprise AI foundation aligned to governance policies

• Compute & AI-Optimized Infrastructure

  Topics
  • EC2 instance families for AI workloads
  • GPU vs Trainium vs Inferentia decision framework
  • Elastic Load Balancing and Auto Scaling

• Subtopics
  • Cost-performance trade-offs
  • Spot vs On-Demand strategy for ML training
  • Placement groups and network throughput
• Hands-on / Lab
  • Launch GPU-backed EC2 instance
  • Benchmark inference workload performance

Real-world application
• Infrastructure sizing strategy for ML training vs inference systems

• Cloud Storage Architecture

  Topics
  • S3 storage classes and lifecycle policies
  • EBS vs EFS decision matrix
  • Data lake design patterns

• Subtopics
  • Encryption at rest and in transit
  • Intelligent tiering for ML datasets
• Hands-on / Lab
  • Build secure S3 data lake with lifecycle rules
  • Implement bucket policies and encryption
• Real-world application
  • Designing scalable storage for training pipelines

• Databases & Vector Data Stores

  Topics
  • RDS vs DynamoDB architecture comparison
  • Vector search fundamentals
  • OpenSearch vector engine & Aurora PostgreSQL pgvector

• Subtopics
  • Indexing strategies
  • Similarity search optimization
• Hands-on / Lab
  • Deploy Aurora PostgreSQL with vector extension
  • Execute embedding similarity queries

Real-world application
• Architecting RAG-ready backend systems

• Advanced VPC Design
  

  Topics
  • Multi-tier VPC architecture
  • NAT gateways, Transit Gateway
  • PrivateLink and secure service access

• Hands-on / Lab
  • Provision production-grade VPC using CloudFormation
• Real-world application
  • Designing isolated ML environments

• Observability & Operational Excellence

  Topics
  • CloudWatch, CloudTrail, X-Ray
  • Centralized logging architecture
  • AI workload monitoring patterns

• Hands-on / Lab
  • Configure monitoring dashboards
  • Implement alerting for endpoint latency

Real-world application
• Production ML system monitoring strategy

• Serverless AI Workloads
  

  Topics
  • Lambda for AI microservices
  • Step Functions for ML orchestration
  • Event-driven architectures

• Hands-on / Lab
  • Build serverless image classification pipeline
• Real-world application
  • Event-driven AI document processing system

• SageMaker Foundations
  

  Topics
  • SageMaker Studio architecture
  • ML lifecycle phases
  • Feature Store fundamentals

• Hands-on / Lab
  • Build exploratory notebook in SageMaker Studio

Real-world application
• Enterprise ML experimentation workflow

• Distributed Training & Hyperparameter Tuning
  

  Topics
  • Managed training jobs
  • HyperPod distributed training
  • Checkpointing and fault tolerance

• Hands-on / Lab
  • Launch distributed training job
• Real-world application
  • Large-scale model training optimization

• Containers & Kubernetes for ML
  

  Topics
  • Docker fundamentals
  • ECR registry
  • EKS for ML workloads

• Hands-on / Lab
  • Containerize ML inference API
  • Deploy to EKS cluster

Real-world application
• Portable ML deployment architecture

• CI/CD for Cloud & ML
  

  Topics
  • CodePipeline automation
  • Infrastructure CI/CD
  • Blue-green deployments

• Hands-on / Lab
  • Implement CI/CD pipeline for ML service

• MLOps with SageMaker
  

  Topics
  • Model registry
  • Pipeline automation
  • Drift detection & monitoring

• Hands-on / Lab
  • Build automated retraining pipeline

Real-world application
• Enterprise MLOps governance model

• Foundation Models & Bedrock Architecture
  

  Topics
  • Bedrock model catalog
  • Amazon Nova, Claude, Llama comparison
  • Inference optimization

• Hands-on / Lab
  • Deploy text generation API via Bedrock
• Real-world application
  • Enterprise GenAI architecture blueprint

• Cost & Security Governance for GenAI
  

  Topics
  • Guardrails and content filtering
  • Token cost management
  • Responsible AI practices

Hands-on / Lab
• Configure Bedrock Guardrails

• Prompt Engineering & LLM Design
  

  Topics
  • Structured prompting
  • Few-shot & chain-of-thought
  • Evaluation frameworks

• Hands-on / Lab
  • Optimize prompts for summarization

• Retrieval-Augmented Generation (RAG)
  

  Topics
  • Knowledge Bases architecture
  • Embeddings & vector search
  • Context window optimization

• Hands-on / Lab
  • Build enterprise RAG chatbot

Real-world application
• Internal policy Q&A AI assistant

• AWS Glue & ETL
  

  Hands-on ETL transformation

• Athena & Analytics

Real-world application
• Building analytics layer for AI insights

• AI APIs (Comprehend & Rekognition)
  

  Hands-on moderation tool

• Agents for Bedrock
  

  Hands-on autonomous agent implementation

Real-world application
• AI-powered workflow automation

• Lex & Conversational Design
  

  Hands-on chatbot

• Enterprise Capstone Project
  

  • Design end-to-end AI architecture
  • Implement secure RAG system
  • CI/CD integrated deployment
  • Monitoring & governance

• Hands-on / Lab
  • Build complete AI solution from ingestion to GenAI interface

Real-world application
• Present enterprise AI transformation blueprint