From code to production: how to keep pipelines running smoothly at scale
In this talk, Yannick and Joshua will share how they designed and evolved the CI/CD pipelines that power a global retailer’s data platform across 30+ countries. From the first setup to today’s architecture, they’ll dive into what worked, what didn’t, and why. Expect real-world lessons around dbt, Terraform, and security – from failed builds to the automation wins that finally made life easier.

This session offers practical insights for anyone working with data: from engineers maintaining pipelines to analysts and ML developers relying on them.

Learnings from integrating AI within CyberBench: the good, the bad and the ugly
In this talk, Lucas shares his journey of building CyberBench: a framework that automatically benchmarks Large Language Models (LLMs) for direct security risks such as prompt injections and data leaks.

After a brief introduction to the cybersecurity risks of GenAI systems and the motivation behind CyberBench, Lucas dives into the technical lessons learned from integrating LLM-based components into a software system. Expect practical insights into building reliable and efficient LLM pipelines and lessons on what worked (and what didn’t).

In short: learn to maximize the good, minimize the bad, and make the ugly a little prettier when building LLM systems in production.

AI agents across domains: from space exploration to energy markets

The common thread in David’s carreer: creating intelligent agents that make critical decisions in real-time. He now applies these principles to energy markets, where agents manage battery systems and navigate complex trading mechanisms including aFRR markets.

In this talk, he’ll reveal the surprising connections between autonomous Mars rovers, supply chain optimization engines, and energy trading agents, showing how the core architecture of perception, planning, and execution remains consistent even as the domains radically change.