Agent-Based AI Systems in Software Architecture
Integration of AI Workflows into Existing IT Environments
.svg)
Machine Learning Engineer
In the “Architectures for Agent-Based AI Systems” training course, you’ll gain practical architectural knowledge to help you understand agent-based AI systems in a structured way and apply them effectively in real-world projects. You’ll learn how agents are conceived as the interaction of core building blocks and what role context, memory, and state play in ensuring reliable behavior. The focus is on interoperability—for example, through standards such as MCP—as well as on quality and observability, and the requirements for security and scalability in production IT environments.
From Model to Agent System
- An overview of what distinguishes agent-based AI systems and why this is changing architectural decisions.
- An overview of the key components that interact within agent-based systems.
Mastering Context, Memory, and State
- The role that context, memory, and state play in ensuring reliable behavior, and where typical pitfalls lie.
- Approaches for managing context in a way that ensures systems remain stable even under real-world conditions.
Tool Integration and Collaboration in Agent Landscapes
- How agents use tools and why interoperability should be considered early on.
- MCP and multi-agent systems as a framework for standardized collaboration and extensibility.
Making Quality Visible
- How quality can be meaningfully conceived and evaluated in non-deterministic systems.
- Observability as the foundation for making operational behavior transparent.
A Smooth Transition to Production
- What matters for productive maturity—from scaling to security—without losing sight of the big picture.
- Guidelines for operations that can be integrated into existing development and release processes.
Requirements:
- A solid understanding of software architecture and the design of software systems, APIs, and integration interfaces
- Basic understanding of GenAI and LLM usage, such as context windows, tokens, and RAG as a concept
Technical requirements:
There are no special requirements for the training environment, as the exercises take place using cloud- and web-based tools such as Miro, Draw.io, and platforms such as Jupyter or HuggingFace. All that is needed is a stable and sufficiently fast internet connection.
- You understand the agentic paradigm and can clearly distinguish agentic AI systems from traditional GenAI scenarios.
- You'll learn how agents are architecturally structured and how their core components can be integrated into a robust overall system.
- You'll learn practical decision-making approaches for context engineering, as well as memory and state management.
- You will learn how tool integration and interoperability are structured in agent architectures, including in the context of MCP and multi-agent systems.
- You will learn the criteria and procedures for evaluating the quality, robustness, and security of agent-based systems.
- You'll learn how to prepare agent-based systems for production environments, with a focus on observability, scalability, and secure operation.
Your advantages at a glance:
- Practical content: The focus is on real-life examples and practical exercises.
- State-of-the-art concepts: You'll gain a structured overview of key approaches in agent-based AI systems and their architectural principles.
- Efficient Integration: A Clear Guide to Integrating and Operating Agent-Based AI Solutions in Existing System Environments.
- Visual collaboration through, for example, whiteboarding
- High level of interactivity through customized exercises and breakout rooms
- Ideal trainer support even in the breakout rooms
Upon completion of the CPSA Certified Professional for Software Architecture Advanced Level® training, you will receive a recognized certificate confirming your expertise and qualifications in the field of software architecture at an advanced level.
Credit points for CPSA-A certification:
With the AGILA training course, you will earn 20 methodological and 10 communication credit points in accordance with the iSAQB Advanced Level Program.
.svg)