Munich Datageeks June Edition

As always, there will be plenty of time for networking, discussions, and snacks & drinks courtesy of our host.
📍 Event Details
📅 Date: Tuesday, June 23rd, 2026

🎤 From Frankensteining DeepSeek to Mixture-of-Experts: The R1T Chimera Models
by Fabian Klemm, TNG Technology Consulting GmbH

At TNG an internal AI research group has been experimenting with Mixture-of-Expert Large Language Models and adaptions of the DeepSeek model family. First they manipulated the way experts work within a model under the name Mixture of Tunable Experts. They then continued with the Assembly-of-Experts merging process resulting in the Chimera models which gained severe attention with daily usage of more than 10 billion tokens via OpenRouter.

🎤 From Models to Systems: How PAYBACK built an AI Platform for Production ML
by Artemii Frolov, PAYBACK & Mohammed Hatem Moustafa, PAYBACK

Building one ML model is hard. Enabling numerous cross-functional teams to ship and operate models reliably is a entirely different game.
This talk shares how PAYBACK built an internal AI Platform on top of Google Cloud to let delivery teams build, deploy, and run ML models in production - without reinventing the pipeline every time. From an engineering angle, we cover how the platform standardizes modular, configurable, and resilient workflows that absorb frequent change instead of breaking under it. From a leadership angle, we show how clear ownership, shared standards, and predictable delivery reduce risk and keep teams aligned as the platform grows.
We then zoom into one concrete use case to make it real: tackling the model retraining stage of the ML lifecycle. You'll see how the platform turns retraining from a manual, error-prone chore into a repeatable, automated capability teams can trust.
If you care about moving from models to systems - and building a foundation that scales as your org does - this talk is for you.

About Artemii Frolov:
Artemii Frolov is a Data Scientist at PAYBACK, where he works on building and scaling machine learning systems for real-world applications. Over the past decade, he has developed end-to-end ML pipelines and production solutions across companies with experience spanning fraud detection, computer vision, and ML platform development. He is particularly interested in turning ML research into reliable, scalable production systems and making training pipelines practical for fast-moving teams.