Munich Datageeks March Edition

We are thrilled to announce our next Meetup on March 26th at Reply.
Format:

• 2 talks (each ca. 40 min incl. discussion)
• Time for networking + food + drinks before, in between, and after the presentations
• Talks are held in English
• We will be taking photos and/or film footage at the event. These will be used to share news about our meetups and to publicize upcoming events.

The lineup:

First talk:
Nico Jahn, Syed Husain Mustafa and Alena Schmickl - From Paper to Insight: Medical Document Processing on AWS with Generative AI

Abstract:
Medical and enterprise workflows often rely on large volumes of unstructured documents, where layout variability and semantic context make reliable extraction difficult. Using a production-ready implementation as our guide, we will demonstrate how AWS’s AI stack provide the foundation for intelligent, multimodal data extraction. We also introduce Kiro, an AI-native developer experience that streamlines the transition from these conceptual architectures to the real-world pipeline featured in this talk.
At the heart of this session is a deep dive into a document processing pipeline designed to transform heterogeneous files into structured, actionable data. The solution combines OCR, document classification, and Large Language Models orchestrated on AWS to handle complex layouts, low-quality scans, and multi-page content. We will walk through the technical architecture and share practical lessons learned when moving from experimentation to a production-ready pipeline, highlighting how AI can significantly reduce manual document review and improve information accessibility.

Bio:
Nico Jahn
I am an AI Solutions Architect and Engineer working as a Senior Consultant at Storm Reply, where I design and deliver production-grade AI systems on AWS for enterprise clients. With over 4 years at Reply and 5+ years of AWS experience, I specialize in taking Generative AI applications from proof-of-concept to production, including intelligent document processing, conversational interfaces, and multi-agent systems. My work spans the full AI development lifecycle, leveraging deep learning, LLMs, and agentic architectures to solve complex business challenges. Beyond AI, I bring broad expertise in AWS networking, compute, and storage, enabling me to design and evaluate end-to-end cloud architectures. In my free time, I run a homelab where I experiment with enterprise-grade servers and networks.
Syed Husain Mustafa
I am an AWS-certified Solutions Architect and Data Engineer working as a Consultant at Storm Reply. Over the past 1.5 years, I’ve delivered Generative AI solutions for enterprise and startup clients, integrating LLM-driven components into production workflows, building computer vision and document intelligence solutions, and facilitating BI migrations to Amazon QuickSight. I engage with customers to shape solution strategy, develop new opportunities, and contribute to internal offerings leveraging AWS and partner technologies. I design scalable, cloud-native systems that turn complex requirements into reliable, end-to-end solutions.
Alena Schmickl
Alena is a Senior Solutions Architect at AWS who has transitioned from the "engine room" of development to the strategic world of cloud architecture. Her professional journey began as a full-stack developer working on complex projects in Computational Linguistics and Knowledge Graphs — technical roots that now inform her pragmatic approach to modern AI. Now at AWS, Alena works at the intersection of startup agility and the rigorous demands of the Healthcare and Life Sciences sectors. From this vantage point, she has a front-row seat to race for innovation in the era of GenAI. Alena’s philosophy is built on the belief that technology should be viewed as a business enabler rather than an end in itself. She maintains this "builder’s mindset" by staying close to the keyboard in her spare time, developing a local gifting app for the Munich community. Whether she is architecting for healthcare startups or coding for local impact, Alena focuses on translating technical complexity into actionable strategies for growth.

Second Talk:
Nicolas Neudeck - Scaling AI Agents With Filesystems and Bash

Abstract:
As Large Language Models evolve into autonomous engineers, the custom tools designed to guide them often turn into bottlenecks. Replacing complex APIs with a standard Bash terminal and filesystem improves agent performance, reliability, and token efficiency. Lessons from superglue, alongside real-world examples from AWS, Anthropic, and Vercel, demonstrate the value of this architectural shift. Allowing the model to directly generate code, SQL, and terminal commands reduces engineering overhead and enables it to solve problems more naturally.

Bio:
Nicolas Neudeck is a Founding Engineer at superglue (YC W25) and a Computer Science graduate from TUM. Formerly with BCG and Amazon, he specializes in scaling Generative AI applications and distributed cloud systems. Nicolas focuses on architectural decision-making that transforms state-of-the-art AI concepts into reliable, high-traffic products.