Uploaded February 2026 | Updated September 2026, 2 weeks ago
Moving a machine learning model from exploration into production is never a straight path.
Academic success often relies on clean datasets, controlled environments, and benchmark metrics—but in the real world, models face data drift, latency constraints, integration challenges, and the risk of wrong predictions directly impacting customers and business processes.
In this talk, we share our journey of bringing fraud detection models from research into production at Swiss Post. We highlight the role of the shadow model approach, where new models run in parallel with production systems to safely validate performance on live traffic.
You’ll learn how shadow testing helps measure robustness, monitor data drift, and align with business KPIs—without introducing operational risk.
Attendees will take away concrete practices for bridging the gap between academic experimentation and real-world operations: how to design safe rollouts, build monitoring pipelines, and decide when a model is ready for prime time.
Moving a machine learning model from exploration into production is never a straight path.
Academic success often relies on clean datasets, controlled environments, and benchmark metrics—but in the real world, models face data drift, latency constraints, integration challenges, and the risk of wrong predictions directly impacting customers and business processes.
In this talk, we share our journey of bringing fraud detection models from research into production at Swiss Post. We highlight the role of the shadow model approach, where new models run in parallel with production systems to safely validate performance on live traffic.
You’ll learn how shadow testing helps measure robustness, monitor data drift, and align with business KPIs—without introducing operational risk.
Attendees will take away concrete practices for bridging the gap between academic experimentation and real-world operations: how to design safe rollouts, build monitoring pipelines, and decide when a model is ready for prime time.
![[VDBUH2026] Andrei Mihai & Bianca Bulbuc - How Agentic AI Empowers Creative Thinking](https://i.ytimg.com/vi/Tl1uqW8RSAo/mqdefault.jpg "[VDBUH2026] Andrei Mihai & Bianca Bulbuc - How Agentic AI Empowers Creative Thinking
As AI systems evolve from reactive tools into autonomous agents, UX Design must adapt to a new paradigm, one where interfaces don’t just respond, but reason, plan, and act.
In this talk, we’ll explore how Agentic AI is reshaping the future of UX, transforming AI from an assistant into a creative and operational partner. Designers will no longer design only screens and flows, but intent, delegation, and collaboration between humans and agents.
We’ll look at how agent-driven interactions change:
how users express goals instead of commands
how interfaces shift from control surfaces to orchestration layers
how designers move from execution to creative direction
The session will also introduce the role of Model Context Protocol (MCP) in enabling agent-based systems to understand context, connect tools, and act consistently across experiences, bridging design intent and autonomous execution.
This is a talk for designers, developers, and product leaders who want to understand what it means to design with agents, for humans.")
![[VDBUH2026] George Patrașcu - Ports, Adapters, and Other Ways to Stop Breaking Your Business Logic](https://i.ytimg.com/vi/UIlFFBZKSaA/mqdefault.jpg "[VDBUH2026] George Patrașcu - Ports, Adapters, and Other Ways to Stop Breaking Your Business Logic
Your business logic shouldn’t care whether it’s talking to a SQL database, a Kafka topic, or a REST API — but in most codebases, it does. In a microservice architecture, this coupling becomes especially painful: as services evolve and API versions change, you end up touching business logic just to keep up with infrastructure churn.
In this talk, we’ll explore Hexagonal Architecture (Ports & Adapters) as a practical answer to this problem.
We’ll start with the “why”: not to dismiss traditional layered architectures — they solve real problems, and it’s the model we use across many of our own services — but to examine the specific caveats that emerge as systems grow.
Where does coupling infrastructure concerns to business logic start to hurt? Hexagonal Architecture doesn’t throw layering out the window; it refines it, offering targeted solutions to those exact pain points. From there, we’ll walk through the core concepts of the pattern — ports, adapters, and the application core.
But this won’t be a textbook walkthrough. Everything you’ll see is grounded in real production experience. We’ll share the challenges we ran into, the mistakes we made, and the tradeoffs we had to navigate — including the pragmatic vs. dogmatic decisions that come up constantly in practice (should your ORM bleed into your domain model? It depends).
You’ll leave with:
– A clear mental model of Hexagonal Architecture and when it’s worth reaching for
– Concrete code examples of ports and adapters in action, written in C#
– Different approaches to structuring your project (and when to choose each)
– An honest look at the tradeoffs — including when not to use this pattern
Whether you’re building a new service or wrestling with an existing one, this talk will give you practical tools to make your business logic more resilient, testable, and independent of the infrastructure around it.")
