Learning Objective:

Deepen your predictive modelling, experimentation and decision-making capabilities. Learn to move beyond model performance and translate analytical outputs into business decisions and measurable action.

Topics:

• Regression strategy
• Regularisation
• Advanced classification
• Classification threshold selection
• Segmentation
• Clustering
• Forecasting
• Recommendation systems
• Anomaly detection
• Risk flagging
• A/B testing
• Experimentation
• Translating model outputs into business action
• Connecting analytical outputs to ROI

Hands-on

• Python

• scikit-learn

• Statistical modelling workflows

• Predictive modelling techniques

• Experimentation and evaluation workflows

[Case-Study]

Predictive Decision Intelligence Solution

Develop a predictive solution and translate its outputs into a business recommendation, decision framework or measurable action.

Learning Objective:

Move beyond model building to understand how AI systems can be explained, monitored and managed in real-world environments. Learn the foundations of reliable, observable and responsible machine learning systems.

Topics:

• Model explainability
• SHAP
• LIME
• Threshold calibration
• Model Registry
• Deployment options
• Batch inference
• Real-time inference
• Monitoring model latency
• Monitoring operational cost
• Model drift
• CloudWatch basics
• Rollback strategies
• Model lifecycle thinking
• Responsible AI foundations

Hands-on

• SHAP

• LIME

• Amazon SageMaker Model Registry

• Amazon CloudWatch

[Case-Study]

Explainable & Monitored ML System

Evaluate a machine learning model, explain its predictions and define a monitoring approach for performance, drift, latency and operational health.

Learning Objective:

Build strong applied Generative AI capability using foundation models and Amazon Bedrock. Learn to design GenAI applications while understanding the cost, performance, security, privacy and governance considerations involved in enterprise adoption.

AWS Self-Paced Content

• Getting Started with Amazon Comprehend

• Amazon Rekognition Getting Started

• Cloud Financial Management: FinOps for GenAI

Topics:

• Foundation models
• Amazon Bedrock
• Bedrock APIs
• Prompt engineering
• Structured outputs
• Summarisation
• Multimodal AI tasks
• Model selection
• Cost and performance trade-offs
• AI governance
• Data privacy
• LLM security
• Model risk

Hands-on

• Amazon Bedrock

• Foundation models

• Bedrock APIs

• Prompt engineering

• Structured output workflows

• Generative AI application workflows

[Case-Study]

Enterprise Generative AI Application

Design and build a Generative AI application using Amazon Bedrock, with consideration for model selection, performance, cost, security and enterprise governance.

Learning Objective:

Learn how to connect foundation models with enterprise knowledge and data using modern retrieval patterns. Build and evaluate Retrieval-Augmented Generation systems for more grounded and useful AI applications.

AWS Self-Paced Content

• Amazon Textract Getting Started

• Amazon Transcribe Getting Started

Topics:

• Embeddings
• Chunking strategies
• Vector search
• OpenSearch concepts
• Kendra concepts
• S3 Vectors
• Bedrock Knowledge Bases
• LangChain
• LlamaIndex
• RAG pipeline design
• Retrieval strategies
• Faithfulness evaluation
• Relevance evaluation
• Prompt-injection testing
• Red-teaming basics
• LLM evaluation

Hands-on

• Amazon Bedrock Knowledge Bases

• S3 Vectors

• Amazon OpenSearch concepts

• LangChain

• LlamaIndex

• Embedding workflows

• Vector search workflows

• RAG pipelines

[Case-Study]

Enterprise Knowledge Assistant

Build a RAG-based knowledge application that retrieves information from enterprise data and generates grounded responses. Evaluate the system for relevance and faithfulness.

Learning Objective:

Move beyond simple prompts and chatbots to design AI systems that can use tools, reason through tasks and coordinate multi-step workflows. Learn the foundations of modern agentic and multi-agent systems.

Topics:

• Agent patterns
• Tool use
• Function calling
• Planner-executor architectures
• Agent memory
• JSON tool schemas
• Model Context Protocol (MCP)
• Agent orchestration
• LangChain orchestration
• Strands
• Multi-agent patterns
• Safety budgets
• Cost ceilings
• Latency ceilings
• Resource governance

Hands-on

• Model Context Protocol (MCP)

• LangChain

• Strands

• Tool-calling workflows

• Agent orchestration patterns

• Multi-agent workflows

[Case-Study]

Tool-Using AI Agent

Design an AI agent that uses tools to complete a multi-step task, with defined controls for cost, latency, safety and resource usage.

Learning Objective:

Learn how to package, deploy and operate AI systems in enterprise environments. Move from a working prototype to a production-oriented system with the security, observability, governance and operational controls required for enterprise use.

AWS Self-Paced Content

• AWS Cost Explorer Getting Started

• AWS Budgets Getting Started

• Introduction to Amazon Virtual Private Cloud (VPC)

Topics

• AI serving architectures

• Bedrock deployment patterns

• SageMaker deployment patterns

• Containerisation overview

• CI/CD pipelines

• Observability

• Traceability

• IAM and security

• Audit logging

• Production guardrails

• Operational runbooks

• Ethics and governance

• Cost management

Hands-on

• Amazon SageMaker

• Amazon Bedrock

• CI/CD workflows

• IAM

• Amazon CloudWatch

• AWS Cost Explorer

• AWS Budgets

• Amazon VPC

[Case-Study]

Production Deployment Blueprint

Create a production deployment approach for an AI system covering architecture, security, observability, cost management, governance and operational processes.

Learning Objective:

Bring together the full program into one employer-relevant, production-oriented AI solution. Integrate data workflows, machine learning, Generative AI, decision support and optional agentic capabilities into a complete end-to-end system on AWS.

Topics

• End-to-end AI solution architecture

• Data workflows using Amazon S3

• Querying with Amazon Athena and Amazon Redshift

• Machine learning with Amazon SageMaker

• Generative AI with Amazon Bedrock

• Decision support with Amazon Quick Sight

• Optional agentic AI workflows

• Monitoring and observability

• Cost estimation

• Governance and responsible AI

• Operational readiness

Hands-on

• Amazon S3

• Amazon Athena

• Amazon Redshift

• Amazon SageMaker

• Amazon Bedrock

• Amazon QuickSight

• Optional agentic AI workflows

[Case-Study]

Production-Ready Data Science & AI Solution on AWS

Build and present an end-to-end AI solution on AWS. The capstone deliverables include a working demo, architecture document, KPI memo, monitoring plan, cost estimate, governance checklist and operational runbook.