Softwarearchitektur für KI-Systeme
SWARC4AI is a three-day iSAQB Advanced Level training on designing software architectures for AI systems, held in person at the entwickler Hub coworking space in Berlin from 9:00 to 17:00 each day. Trainer Jan-Torben Evers covers how AI components differ from classical software and how to integrate machine learning and generative AI into existing systems as hybrid solutions that stay scalable, secure and explainable. The agenda runs through compliance and security topics including the EU AI Act, data protection, adversarial attacks and data poisoning; design patterns for AI systems and their integration into existing architectures; data pipelines and data engineering tooling; and operational quality attributes such as monitoring, resource efficiency and MLOps. A dedicated block deals with system architectures for generative AI, including LLM tooling patterns and retrieval-augmented generation. Theory alternates with hands-on exercises and case studies that apply the concepts to realistic project scenarios. Participants are expected to arrive with a working grounding in machine learning, data processing and software architecture, and the module counts for 10 methodical and 20 technical iSAQB credit points.
