From AI demos to production reality
The Munich AI Nexus is a meetup series for experienced architects and engineers who are tackling real-world challenges in deploying AI systems into production. The appliedAI Developers bring together engineers, researchers, and technology leaders to exchange knowledge, develop solutions, and advance practical AI together.
Why does this meetup exist?
AI meetups often focus on impressive demos – while the challenges and failures that emerge in real-world production environments receive far less attention.
For example:
- Why does a model work perfectly in a test environment but crash in production at 2 a.m.?
- Why does a data pipeline end up being held together by temporary fixes and workarounds?
- Which architectural decisions help prevent AI systems from becoming unreliable?
This meetup series addresses exactly these gaps. Across nine sessions, it explores the topics that become genuinely challenging when building and operating real-world AI systems.
The result: the Production Readiness Blueprint– an open-source collection of architectural patterns and practical experiences that have already helped teams successfully bring AI systems into production.
This session is hosted by fortiss. In addition to insights into the fortiss labs and a demo tour on neuromorphic computing with Michael Neumeier (Competence Field Neuromorphic Computing), participants can look forward to a talk by Dr. Yuanting Liu (fortiss) on how to design AI systems that keep humans genuinely in control.
Overview
-
As AI systems become capable of making decisions autonomously, processing information, and executing tasks, one central question emerges:
How do we keep humans in control?
This session explores the collaboration between humans and machines in production AI systems and shows how human responsibility, oversight, and feedback can be meaningfully integrated:
- When should a human confirm, intervene, or override a decision?
- How can an AI system recognize uncertainty and communicate it transparently?
- How can responsibilities between users, operators, and AI agents be clearly defined?
- How can human feedback become an effective part of the system?
- How can control and safety be ensured without unnecessarily slowing down processes?
Your experiences help make practical knowledge visible:
- Challenges from your AI projects
- Proven approaches and lessons learned
- Problems, solutions, and experiences from real-world deployments
-
- Demo tour of fortiss Labs and insights into neuromorphic computing
Michael Neumeier, Research Scientist in the Neuromorphic Computing Competence Field
- Talk: How do we build AI systems that keep humans truly in control?
Dr. Yuanting Liu (fortiss), Head of Human-Centered Engineering at fortiss and author of Human-Centered Machine Learning
- Group Challenge
We will explore a simplified multi-step AI workflow: it works under ideal conditions – but fails in real-world scenarios.
The challenge: Instead of fixing the code, design a state-layer architecture that makes the system more robust. (Details to follow.)
- Networking & Pizza
- Demo tour of fortiss Labs and insights into neuromorphic computing
-
Location
fortiss GmbH
Guerickestr. 25
80805 MunichLanguage
EnglishParticipation fee
Free of charge
-
- Curator: Asaad Almutareb
Founder of artiquare and curator of the appliedAI Developers MUC Meetup - AppliedAI Developers
- Host & venue: fortiss
- Curator: Asaad Almutareb