Software Architecture for AI Systems

Setting up AI systems with a sound architecture (3 days)

Software Architecture for AI Systems: Setting Up AI Systems with a Sound Architecture

From fundamental concepts like Machine Learning, AI Engineering, LLMs, Foundation Models, Prompting or RAG, through specific quality challenges for AI systems, to designing complex agentic systems: our seminar covers all architecture-relevant facets for designing AI-based applications and application landscapes.

Agenda

Introduction and Motivation
When and where does using AI pay off? What’s actually the difference between Machine Learning and AI Engineering? What specific challenges does developing AI-based systems bring, and where exactly do the differences lie compared to developing classic, deterministic systems?

Quality Attributes for AI Systems
Understanding and formulating AI-specific quality requirements

AI Technologies
Assessing the use of symbolic AI, Machine Learning, or Foundation Models

FTI Pipelines for Machine Learning
Setting up feature, training and inference pipelines for Machine Learning

Foundation Models
Understanding how LLMs work

Use Cases for AI Engineering
Which use cases can be solved based on Foundation Models?

Prompting, RAG, Fine-Tuning and Distilling
Knowing the fundamental techniques for working with Foundation Models

Evaluating AI Systems
How can the results of AI systems, particularly generative AI, be assessed?

Agentic AI
Understanding and designing agentic systems

Your Benefit

After this seminar, you can soundly assess AI initiatives, scope them precisely, and translate them into viable software architectures. You recognize when AI delivers real value, address relevant quality goals, and learn to make practical architecture decisions for AI systems. You’re able to robustly design data flows and pipelines from feature engineering to inference, position LLM-based approaches against classic ML or symbolic methods, and select suitable integration patterns. You also know how to equip agents with the right tools, orchestrate them, measure results, and evaluate risks.

Target Audience

This workshop is aimed at software developers and architects who design and evolve systems and are thinking about the use of AI in their software. You already have practical experience in IT projects. Knowledge of a specific technology or programming language is not required.

Our Extras for You

Photo Log, Slides and Handouts
All exercise results, handouts, and theoretical content shown are made available to participants electronically afterwards

Additionally...
You’ll find a quick, practical overview of Machine Learning in our architecture cheat sheet no. 10 Machine Learning Lösungen entwerfen (German)

Architekturspicker Nr. 10 Machine Learning Lösungen entwerfen

Did we spark your interest?

Verwenden Sie das Kontaktformular oder senden Sie Felix Kammerlander eine E-Mail.