Content
- 01
Introductionthe software architecture concepts that matter for AI, including the vocabulary, possible use cases, risks, strengths and weaknesses, and when you're better off with AI or with classic software development.
- 02
Compliance, security and alignmentthe legal and regulatory landscape of AI systems that software architects need to know, including EU AI Act compliance, data protection law and risk classification frameworks.
Technical security strategies to protect AI systems against attacks, robust data management practices, and proper documentation for traceability and transparency.
Ethical AI governance: safety principles, responsible AI frameworks, and how to build compliance, security and ethics into your architecture decisions.
- 03
Understanding ML systemsthe basics of machine learning project lifecycles and software development processes for AI systems. Defining ML tasks, understanding the problem's requirements, and specifying the input and output data of ML systems.
- 04
Exploratory data analysisdifferent data types and typical ML problems, and how data scientists acquire, label and analyse datasets. How to spot bias in data, and the storage options for ML projects.
- 05
Data cleansing and preparationstrategies for data aggregation, cleansing, transformation and augmentation with modern data engineering tools. How data scientists design efficient data pipelines and handle the data quality problems that can hurt ML systems.
- 06
Feature engineeringhow to prepare input data for different AI algorithms, and the trade-offs between model architectures. Advanced techniques such as transfer learning and fine-tuning to optimise model performance.
- 07
Model training and model selectionthe hardware requirements for training models, and how to track model parameters, metrics and results. Model evaluation techniques that weigh cost, sustainability and reproducibility.
- 08
Building a proof of conceptchoosing the right tools to prototype AI systems and applying design patterns for ML applications. User interface design principles and the explainability requirements of AI systems.
- 09
Deploying ML systemsCI/CD pipelines, deployment strategies and platform options for ML models, including SaaS and self-hosted solutions. Integrating ML models into existing systems while keeping security, compliance and deployment architecture in mind.
- 10
Operating ML systemsthe monitoring needs specific to AI systems, and measuring performance metrics such as latency and throughput. Strategies for scalability, robustness, reliability and fault tolerance in production ML systems.
- 11
Maintaining ML systems over timespotting and handling the different kinds of model drift and ongoing data quality problems. Using user feedback to improve the model and keep the system performing over the long run.
- 12
System architectures and platforms for generative AIhow to integrate AI systems, especially large language models (LLMs), into real IT architectures with practical tools, libraries and interfaces.
The basics of generative AI and LLMs, together with techniques such as prompt engineering, retrieval-augmented generation (RAG) and agentic workflows, to build effective AI applications.
The practical side too: system evaluation, cost management, quality assurance and design patterns for running AI solutions in production.