AI at scale hosted by Zalando

17:30-18:00

• Doors open
• Welcome words by Zalando & WiMLDS

18:00- 18:40
'Deploying AI agents at scale' by Valentina Halasi
Abstract:
You've built AI agents with a framework like CrewAI or LangGraph. They work great locally. Now what?
This talk covers the real infrastructure challenges of deploying multi-agent systems: handling long-running executions, managing stateful workflows, versioning deployments, and debugging complex agent behavior in production. Learn the steps from "works on my laptop" to production-ready, and discover why most teams end up building the same infrastructure stack.
I'll wrap up with a live demo showing how to deploy agents with a single command, stream execution in real-time, manage secrets securely, auto-scale to handle concurrent runs, and version deployments with instant rollbacks, without building the full infrastructure yourself.

18:45-19:30
'AI at scale in practice: Zalando Assistant' by Katya Mirylenka
Abstract:
As artificial intelligence transforms the fashion industry, multimodal AI agents are stepping up to deliver deeply personalized shopping experiences. Using the Zalando Assistant (ZA) as a real-world case study, this talk dives into the practicalities of building and operating these complex systems at scale. We will explore why the e-commerce industry is moving beyond simple chatbots to sophisticated LLM-powered agents and how ZA processes complex user interactions to elevate the customer journey. Finally, we will unpack the behind-the-scenes technical hurdles of large-scale AI, including optimizing search and retrieval under heavy load, refining LLM evaluations, and handling complex product data.

19:30-20:30
Networking & Mingling