The program at a glance
The CAS LLMs and AI Agents (bootcamp) program fills the gap between prototype and reliable operation using technical, application-focused and state-of-the-art language AI: Large language models, AI agents and their integration in real-life work processes, data landscapes and existing IT systems. This CAS teaches participants to systematically design, realize, evaluate, secure, deploy and operate LLM-based applications and agent systems.
Learning objectives
In diesem CAS werden Sie zur Fachperson, die LLM-basierte Anwendungen und AI-Agent-Systeme konzipiert, evaluiert, absichert und betreibt – und in Ihrer Organisation Verantwortung für deren sicheren Einsatz übernimmt.
After completing the CAS, you will be able to
- identify, evaluate and define suitable use cases for LLMs, retrieval-augmented generation (RAG) and AI agents and select the best technical solution;
- develop AI applications that access reliable data and knowledge sources;
- link AI agents with tools, data, and existing systems;
- evaluate the quality, security, costs and speed of AI solutions;
- address risks such as hallucinations, data loss or unwanted system interactions;
- install, monitor and continually improve AI solutions.
Throughout the CAS program, you will develop a transfer project taken from your own professional practice.
What the program offers:
- Practice projects: You will build, evaluate and operate your own LLM and agent applications and develop a portfolio from the first working application to a virtually production-ready solution.
- Learning with and from experts: Experienced professionals will guide you through the modules, offer personal feedback and advise you on conceptual and technical issues.
- Production readiness as a core skill: You will not only learn how LLM applications and AI agents work, but also how to assess their quality, control the associated risks and use the technology responsibly.
- Transfer-oriented project work: We integrate real-life use cases such as company-specific micro projects and exploratory proof-of-concepts. Your assessed assignment is a project of your own choosing, which you will develop across the six modules and independently complete with optional coaching.
- Career boost: If your goal is to lead the implementation of LLM-based solutions or assume responsibility for them, this CAS will give you the necessary skills.
- Contacts and networking: You will connect with specialists and managers from a wide range of industries and benefit from the shared learning experience in a highly motivated community.
Modules and contents
The CAS is a 12-day program with six two-day modules on Fridays and Saturdays. Each module starts by building a directly applicable core skill (day 1). In a second step, it then offers more in-depth insight into quality, security, integration, deployment and operation (day 2).
- Day 1 (Friday) – Basics and hands-on application: The first day of the module builds comprehensive and directly applicable core skills around LLMs and AI agents.
- Day 2 (Saturday) – Focus on productive operation: The second part of the module discusses Friday’s topic in a real-life context: Quality of results, security, integration in existing systems and data landscapes, deployment and operation.
Module 1: Foundations of LLMs and AI System Design
- Day 1 – Basics: In-depth introduction to the key concepts of natural language processing and large language models: tokenization, vector representations, transformer models and prompt engineering. You will apply them in realistic exercises and build your first structured LLM application.
- Day 2 – Focus: Designing reliable LLM applications: Structured model responses, defined inputs, expected results, and known limitations. You will develop your first stand-alone application as a starting point for your transfer project.
Module 2: Knowledge Systems and Retrieval-Augmented Generation (RAG)
- Day 1 – Basics: How LLM applications access documents, internal information and knowledge sources. You will build retrieval-augmented generation pipelines for context optimization and verifiable responses.
- Day 2 – Focus: You will improve the quality of the information retrieved and work hands-on with confidential data, access rights, and the maintenance of knowledge sources – prerequisites for robust RAG systems.
Module 3: AI Agents and Controlled Workflows
- Day 1 – Basics: Mastering the difference between chatbots, automated workflows and AI agents. You will build agent systems that independently structure tasks, control tools and make decisions, and learn the model context protocol (MCP) for consistent access to external tools, data sources, and APIs. Other topics include agent orchestration and multi-agent workflows, ranging from sequential task processing to parallel agent systems.
- Day 2 – Focus: You will build a controlled agent that processes information and interacts with a tool or external system according to set rules. The focus is on clear processes, authorizations, approvals and transparent decision-making – the conditions under which agents take on real-world tasks.
Module 4: Quality, Safety and Responsible Use
- Day 1 – Basics: Why performance alone is not enough: Bias detection, fairness and transparency in LLM-based systems. You will use red-teaming strategies to test LLMs for vulnerabilities, undesired output and safety risks.,
- Day 2 – Focus: You will systematically evaluate your LLM applications: Test cases, challenging inputs, quality criteria, response quality. You will address key risks including hallucinations, manipulation attempts (jailbreaks, prompt injections), data loss, unwanted agent actions, and implement protective measures such as guardrails, approvals, and clear rules for the use of AI.
Module 5: Customizing, Deploying and Scaling Models
- Day 1 – Basics: When does it make sense to use existing models and when should you use additional sources of knowledge, make adjustments, and build smaller, specialized models? You will weigh the effort required, the quality, data protection and potential benefits, and test the options in practice.
- Day 2 – Focus: You will explore different forms of deployment – cloud, on-premises, hybrid – and their effect on costs, speed, data protection, and integration into existing IT landscapes. You will then deploy your own application.
Module 6: LLMOps and Operating Production-Grade AI Systems
- Day 1 – Basics: How LLM applications are packaged technically, evaluated, and released: You will work hands-on with automated evaluation, versioning, interfaces, logging, and monitoring.
- Day 2 – Focus: Operations after go-live: You will monitor the quality, costs and system performance, address errors and outages, and continuously improve production operations. You will finalize the operational plan for your transfer project.
