AI SOLUTIONS ARCHITECT

Meet your instructor, get your IDE and source control set up, and see how four cumulative assignments build towards a deployed, production-ready capstone project.

• Meet your instructor
• Course structure
• IDE and source code management set-up
• Assignments and final project overview

Discover who AI Solutions Architects are, where they sit in the modern organisation, and how the role and the teams around it have changed over the last five years.

• The AI Architect role
• Position within the organisation
• How the architect role has evolved
• How AI is reshaping teams
• Workshop: Building a sample AI architecture with an AI coding assistant configured for diagrams

Explore the mindset shift that comes with adopting AI, from the rise of the renaissance developer to a clear-eyed look at what is real and what is hype today.

• Renaissance developers and the changing craft
• The AI landscape: hype vs reality
• Case study: How leading companies built AI architecture capabilities

Learn how the pieces of the generative AI stack fit together, then implement a working RAG pipeline over a prepared dataset and test how well it retrieves.

• Parsing and chunking strategies
• Vector databases and similarity search
• Retrieval tuning and quality evaluation
• Prompt engineering in practice
• Demo: A working RAG system from query to grounded response
• Workshop: Implementing a RAG pipeline over a pre-prepared dataset

Unpack how LLMs work, from tokenisation and context windows to embeddings, and why hallucination makes retrieval essential, before examining generative AI use cases from the pharmaceutical industry.

• LLM architecture, training, and inference
• Hallucination and grounding
• Embeddings and vector representations
• Prompt-engineering theory
• Case study: Generative AI use cases in the pharmaceutical industry

Assignment #1: Multi-Modal RAG System
Build and evaluate a RAG system over a multimodal dataset that returns grounded, cited responses. It becomes the retrieval component for Assignment #2 and the capstone.

Master the anatomy of an AI agent, then build one from scratch with a system prompt, tools, guardrails, its own MCP server, and both short and long-term memory.

• System prompts, tools, and guardrails
• Building an MCP server
• Short and long-term memory
• Evaluating agent performance
• Demo: Observing an agent's reasoning, tool selection, and multi-step task completion
• Workshop: Building an AI agent from scratch

Examine what makes an agent an agent, compare the core architecture patterns, and learn how memory systems and guardrails keep autonomous behaviour safe and under control.

• Agentic loops and planning
• ReAct, Plan-and-Execute, and Reflection patterns
• Memory systems
• Guardrails and human-in-the-loop design
• Safe sandbox environments

Assignment #2: Agentic Application Using RAG as a Tool
Wrap your Assignment #1 RAG system as a tool inside a single reasoning agent that decides when to retrieve, uses memory, and completes multi-step tasks.

See an orchestrator delegate to specialist agents on a clinical research task, then build your own multi-agent system with shared memory, task routing, and full tracing.

• Multi-agent orchestration patterns
• Inter-agent communication and shared state
• Task decomposition and routing
• Combining small and large language models
• Demo: Biomedical research agent with specialist sub-agents
• Workshop: Building a complete multi-agent system

Learn when to choose hierarchical, swarm, or supervisor orchestration, and how evaluation-driven development and reliability engineering keep agents dependable at enterprise scale.

• Hierarchical, swarm, and supervisor patterns
• The agentic loop in depth
• Test harnesses and eval sets
• Retries, fallbacks, and circuit breakers
• Case study: Enterprise multi-agent deployments and lessons learned

Assignment #3: Multi-Agent System
Extend your Assignment #2 agent into an orchestrator with specialist agents, adding evaluation, observability, and memory. This is the system you'll deploy to the cloud for your capstone.

Explore how the AI-Driven Development Lifecycle replaces traditional SDLC for non-deterministic systems, and how it reshapes team roles, stakeholder communication, and the business case for AI.

• From Waterfall and Agile to AI-DLC
• Why traditional SDLC breaks down for AI
• The AI-DLC lifecycle
• Team roles and stakeholder communication
• The business case for AI-DLC
• Case study: An AI-DLC sprint from business problem to deployed solution in two weeks

Gain a practical framework for deciding which AI workloads to fund, scale, or retire, linking token spend to business outcomes through credible, repeatable evaluation methods.

• Why AI economics differ from traditional software
• Budgeting and model routing
• The Frame, Build, Measure, Validate, (de)scale framework
• Linking token spend to business outcomes
• Evals, LLM-as-judge, and regression testing
• Case study: A fund, scale, or retire decision on a real GenAI deployment

Work through a guided Amazon Bedrock AgentCore notebook to deploy a multi-agent workflow on AWS, with identity, gateway controls, observability, evaluation, and guardrails wired in.

• Secure, scoped agent identity
• Centralised gateway controls
• Observability and evaluation at deploy time
• Prompt-injection defences
• Mapping the pattern to other cloud providers
• Cost and latency in production

Find out why so many agentic projects stall at proof of concept, and how designing observability, evaluation, and guardrails in from day one closes that gap for good.

• Why agentic projects fail at POC
• What to instrument in an agentic system
• Offline and online evaluation
• Guardrails by design
• Managed agent runtimes, registries, and harnesses

Assignment #4: Production-Ready Capstone Project 
Bring your multi-agent system to a production-ready state with identity, evaluations, observability, and guardrails in place.

Present your capstone to the class, translate your AI solution for a business audience, and sharpen it with detailed feedback from your peers and instructor.

• Presenting AI solutions to business audiences
• Capstone presentations
• Peer and instructor feedback

Hear senior leaders discuss AI transformation, the evolving AI Solutions Architect role, and where trust, risk, and opportunity meet as organisations adopt agentic AI at scale.

• Vision and AI transformation journeys
• The evolving AI Solutions Architect role
• Short-term vs long-term gains of agentic AI
• Concerns, opportunities, and trust in AI
• Course wrap-up