Speakers:
From Monolith to Modular: How RTL Scaled Its Video ML Pipelines with Python Components and Kubeflow
Date:
Tuesday, November 18, 2025
Time:
2:35 pm
Track:
Room:
Spandau
Summary:
Discover how RTL transformed its video ML pipelines by moving from a monolithic setup to modular Python components using Kubeflow. This session shares real-world insights into building scalable, collaborative workflows across engineering and data science. Learn how we integrated evaluation and metadata tracking to boost agility, transparency, and product quality in a legally restricted environment.
Speakers:
Safeguarding Online Marketplaces: Building a Robust AI-Powered Fraud Detection System for Mobile.de
Date:
Tuesday, November 18, 2025
Time:
3:05 pm
Track:
Summary:
Online marketplaces face growing fraud risks hidden in user conversations, not just listings. At Mobile.de, Germany’s largest car marketplace, we built an AI-driven system to detect fraud using language models, behavioral signals, and metadata. This session shares our journey, key challenges, and practical insights. Takeaways include how to build scalable, adaptive fraud detection to protect trust in real time.