AI SOLUTION ARCHITECT

Get the lay of the land, meet your instructor, and see what’s ahead. We’ll kick things off with where AI stands today, where it’s headed next, and why there’s never been a better time to dive into tech.

• Instructor intro
• Course structure
• Assignments & final project
• AI overview 

Trace AI’s journey from rule-based systems to GenAI, decode the jargon, and see how architects help shape the big picture through real-world case studies.

• A brief history of intelligent systems
• Case Studies: ELIZA, Chess (IBM Deep Blue), Go (Google DeepMind AlphaGo)
• Core concepts of traditional ML frameworks
• Terminology
• Roles in AI projects & Cloud Solution Architect 

Self-Study Activity: Pick one of the case studies (ELIZA, Deep Blue, AlphaGo, ChatGPT) and write a short reflection: What problem was it solving? Why was it a breakthrough?

Get clear on what makes GenAI tick and how it’s not your regular ML. Learn the key models, the lingo, and when to call in GenAI vs. stick with traditional ML.

• Case Study: ChatGPT
• Types of GenAI models
• Terminology
• Traditional ML vs. GenAI
• Why and when to use ML & GenAI

Self-Study Activity: Write a short reflection on the ChatGPT Case Study: What problem was it solving? Why was it a breakthrough?

Assignment #1: Choose an industry of interest and identify three AI use cases. For each, decide whether it can be addressed with Traditional ML, Generative AI, or a combination of both, and briefly explain your reasoning.

Build your GenAI foundation: understand how leading models work, who’s building them, and get your setup ready to run them yourself.

• Case study: ChatGPT
• Workshop: Setting up the environment
• GenAI models: Core concepts
• GenAI model providers & open source options
• Leading cloud providers

Self-Study Activity: Deploy a GenerativeAI model and call its API from the cloud platform.

Craft prompts that get the results you want. Experiment with zero-shot, few-shot, and chain-of-thought techniques, fine-tune outputs, and practice hands-on in a workshop to see which prompts perform best.

• Zero-shot, few-shot, and chain-of-thought prompting
• Structured output prompts & parameter tuning
• Evaluating & iterating prompts for better accuracy
• Workshop: Applying prompt engineering techniques

Assignment #2: Design and implement two prompts for a real industry use case: one simplistic and one well-engineered. Test both prompts across two different foundation models, compare the outputs, and analyse how prompt quality and model choice impact performance.

Get hands-on with generative AI: explore fine-tuning vs. RAG, understand when to use each, and build your first working RAG pipeline from scratch.

• Introduction to fine-tuning
• Retrieval-augmented generation (RAG)
• RAG vs. fine-tuning
• Workshop: Implementing a basic RAG pipeline

Self-Study Activity: If you had a startup with limited budget, would you choose RAG or fine-tuning for your first AI product? Consider data freshness, compute cost, and ease of updates.

Learn to prep data like a pro: chunk, embed, and store it in vector databases to power retrieval-augmented AI workflows — then put it into practice in a hands-on workshop.

• Chunking strategies & why they matter
• Embeddings & semantic search
• Vector databases
• Workshop: Ingesting & storing a dataset in a vector DB

Self-Study Activity: Create a power point presentation where you create a proposal for a solution with a RAG architecture.

Assignment #3: Implement the RAG architecture with working code. Ingest + chunk + embed a dataset, store in vector DB, build retrieval + generation flow. Explain and justify your chunking strategy of choice.

Explore real-world AI misuse, understand global and local regulations, and apply responsible AI frameworks to design ethical, trustworthy, and compliant solutions.

• Case Studies: Global & local examples of AI misuse
• Law & governance: Global & local regulations
• Responsible AI frameworks: Traditional & GenAI

Self-Study Activity: Work with your industry scenario, applying RAG to create a solution for your industry. What databases or data layers would you connect to? Map out 3 possible Responsible AI risks & propose one mitigation for each.

Bring your GenAI ideas to life. Combine everything from Weeks 1–4 into a working RAG-based prototype and learn the best practices that make your POC practical and scalable.

• Creating a working GenAI solution
• RAG POC architecture walk-through
• Best practices: What makes a good GenAI POC
• Workshop: Assembling a working POC in code 

Assignment #4: Start your Capstone Project. Design and implement a working Proof of Concept (POC) for a GenAI solution that integrates RAG architecture, vector databases, embeddings, and prompt engineering. Demonstrate your chunking strategy, identify Responsible AI risks with mitigations, and test your solution with sample queries.

Explore the evolution from simple GenAI calls to autonomous agentic systems. Identify patterns, value, and common pitfalls across industries.

• Agentic AI explained: reasoning, planning, acting
• Common patterns of Agentic AI solutions
• Real use case examples
• Industry agnostic examples
• Public & Private sector examples

Dive into what differentiates an agent from a simple LLM call: memory, reasoning, planning, and tool integration. Apply these concepts in a hands-on workshop.

• Core components of a single-agent system
• Agentic memory and context management
• Tool integration & reasoning pipelines
• Common challenges: hallucination, looping, guardrails

Learn to coordinate multiple agents to work in harmony, handle inter-agent communication, and manage tasks efficiently.

• The art of task decompositions
• Collaboration patterns
• A2A: Agent to agent communication
• Challenges

Homework (Individual): Design how you would evolve your RAG POC architecture into an Agentic Solution (either a single agent or a multi -agent architecture) , for a specific use case of your choice.

Learn strategies to balance cost, speed, and output quality while preparing your solution for production.

• Divide and conquer for performance optimisation
• Balance cost, quality and speed
• Agentic optimisation

Homework (Individual): Estimate monthly costs and latency for your Agentic Solution and propose optimisation strategies.

Learn how to move from prototype to production, ensuring reliability and maintainability of your agentic solution.

• MLOps vs GenAI Ops vs GenAI Ops
• GenAI Ops/Agent Ops maturity levels
• Best practices from dev → test → prod 

Homework (Individual): Consider what you will need to implement in a GenAI Ops / Agent Ops workflow for your Agentic Solution (conceptual only).

Learn how to monitor, maintain, and continuously improve GenAI solutions to ensure reliability, compliance, and business value over time.

• Technical challenges
• Business challenges

Homework (Individual): Consider a way to monitor and alert the system owners for your Agentic solution, post-go-live (conceptual only).

Present your working POC, architecture, and optimisation strategies. Engage in peer and instructor reviews to refine your approach.

• Review capstone projects
• Peer and instructor feedback

Clarify career paths, market demand, and emerging trends in AI architecture. Build a roadmap for continued growth.

• Role clarity
• Skills in demand
• Building your profile
• Interview prep
• Future AI trends

Self Study Activity: Create a personal action plan for your GenAI career path.