Amazon Bedrock
This course covers Amazon Bedrock from fundamentals through advanced production architecture. Learn to select and invoke foundation models, engineer prompts, build RAG applications and Bedrock Agents, configure Guardrails, evaluate and customize models, integrate AWS services, and deploy secure, observable, cost-aware generative AI solutions.
Curriculum Areas
- Module 1 - Generative AI & AWS Fundamentals
- Module 2 - Amazon Bedrock Foundation Models
- Module 3 - Amazon Bedrock Console
- Module 4 - Prompt Engineering
- Module 5 - Bedrock API & SDK
- Module 6 - Amazon Bedrock Knowledge Bases
- Module 7 - Bedrock Agents
- Module 8 - Bedrock Guardrails
- Module 9 - Prompt Management
- Module 10 - Model Evaluation
- Module 11 - Bedrock Fine-Tuning & Customization
- Module 12 - Embeddings & Vector Search
- Module 13 - Bedrock with AWS Services
- Module 14 - Security
- Module 15 - Monitoring & Cost Optimization
- Module 16 - Advanced Bedrock
- Hands-On Projects
Amazon Bedrock Curriculum
- Generative AI fundamentals
- LLMs and Foundation Models
- Generative AI vs traditional AI/ML
- AWS AI/ML ecosystem
- Amazon Bedrock overview
- Bedrock architecture and use cases
- Foundation Models (FMs)
- Model providers available through Bedrock
- Model selection
- Model capabilities and limitations
- Text generation
- Conversational AI
- Embeddings
- Multimodal AI
- Model parameters and inference settings
- Creating and configuring Bedrock resources
- Model access
- Bedrock playground
- Testing foundation models
- Prompt experimentation
- Inference parameters
- Understanding model responses
- Cost considerations
- Prompt engineering fundamentals
- Zero-shot prompting
- Few-shot prompting
- Role-based prompting
- Instruction prompts
- Context engineering
- Structured outputs
- Prompt optimization
- Reducing hallucinations
- Amazon Bedrock APIs
- Runtime APIs
- Python + Boto3
- Calling foundation models programmatically
- Request and response formats
- Streaming responses
- Error handling
- API authentication
- Introduction to RAG
- RAG architecture
- Data sources
- Document ingestion
- Chunking
- Embeddings
- Vector databases
- Retrieval
- Knowledge Base configuration
- Querying enterprise documents
- AI Agent fundamentals
- Bedrock Agents architecture
- Creating an Agent
- Agent instructions
- Action Groups
- Lambda integration
- Knowledge Base integration
- Agent orchestration
- Agent testing
- Agent deployment
- Responsible AI
- Guardrails architecture
- Content filtering
- Denied topics
- Sensitive information protection
- Prompt attack protection
- Grounding checks
- Configuring Guardrails
- Testing Guardrails
- Creating reusable prompts
- Prompt templates
- Variables
- Prompt versions
- Prompt optimization
- Managing prompts across applications
- Foundation model evaluation
- Automated evaluation
- Human evaluation
- Accuracy and relevance
- Toxicity and safety
- RAG evaluation
- Comparing models
- Performance metrics
- When to customize a model
- Fine-tuning concepts
- Training datasets
- Validation datasets
- Model customization workflow
- Hyperparameters
- Evaluation of customized models
- What are embeddings?
- Text embeddings
- Semantic search
- Similarity search
- Vector stores
- Amazon OpenSearch
- Aurora PostgreSQL
- Metadata filtering
- Amazon S3
- AWS Lambda
- API Gateway
- Amazon OpenSearch
- DynamoDB
- CloudWatch
- IAM
- Step Functions
- Event-driven GenAI applications
- IAM for Amazon Bedrock
- Roles and policies
- Least-privilege access
- Encryption
- Data protection
- Private connectivity concepts
- Secrets management
- Audit and compliance
- CloudWatch monitoring
- Logging
- Model invocation monitoring
- Token usage
- Latency
- Cost optimization
- Model selection for cost/performance
- Production troubleshooting
- Agentic AI
- Multi-agent architectures
- Tool calling
- Function calling
- Agent orchestration
- Human-in-the-loop
- Multimodal applications
- Enterprise GenAI architecture
Project 1 - Enterprise RAG Chatbot
- S3 -> Knowledge Base -> Embeddings -> Vector Store -> Bedrock FM -> Chatbot
Project 2 - Customer Support Agent
- Bedrock Agent
- Knowledge Base
- Lambda Action Group
- Customer database
- Automated responses
- Human escalation