Foundation Models in Fintech @Klarna & Sovereign AI @Iliad

Hello PyData People!
We are excited to announce our second event of 2026! This time, we will be hosted at iliad Italia's headquarters in Milan for an evening dedicated to AI at scale—ranging from Foundation Models for credit risk to sovereign and open-source data architectures.
📅 When: Wednesday, February 18, 2026
📍 Where: iliad Italia, Viale Francesco Restelli, 1A, 20124 · Milano
⚠️ Important: Spots are limited. Please keep your RSVP updated to allow others to participate if you can no longer attend.

***

## 🕒 Agenda

• 18:30: 🚪 Doors open & Check-in
• 19:00: 🎤 Talk 1: Open Source & Sovereign Architecture – Riccardo Lorenzon
• 19:45: 🎤 Talk 2: Klarna’s Foundation Model – Valeria Verzi, Sudhakar Periyasamy, Goran Dizdarevic
• 20:30: 🍕 Networking & Aperitivo offered by Iliad!

***

## 🎤 The Talks

## 1️⃣An Open Source and Sovereign Architecture for Data, ML and AI

Speaker: Riccardo Lorenzon (Head of Data @ iliad Italia)
Iliad Group invests billions in sovereign AI ecosystems, open-source tech, and low-carbon data centers to reduce reliance on non-European providers and bolster Europe's technological autonomy. True to Iliad Group's 'make-over-buy' culture of full control, iliad Italia is building a data architecture in a hybrid on-prem and Scaleway (Iliad Group cloud) setup, combining open-source tech and managed services.
This powers data management, self-service analytics, and ML/AI use cases for our 12m+ user base and 1000+ employees. During this speech, we will guide you through the tech journeys we have built and are building.

About the Speaker:
Riccardo Lorenzon serves as Head of Data at iliad Italia, where he works on evolving the data function into a modern setup covering the full data value chain from sources to business decisions. He previously led data teams at Vedrai, an AI startup, and Whirlpool EMEA, handling projects in analytics, machine learning, and cloud platforms across areas like consumer service, supply chain, and sales. With a Master’s in Mathematical Engineering from Politecnico di Milano and studies at ETH Zürich, he combines a hands-on approach to delivery with a strong focus on strategic value for the business.

***

## 2️⃣ Building the Klarna Foundation Model: A Unified Transformer for Customer Behavior Modeling at Scale

Speaker: Valeria Verzi (Principal Data Scientist @ Klarna), Sudhakar Periyasamy, Goran Dizdarevic
In this session, we introduce Klarna’s Foundation Model (KFM): a lightweight Transformer trained on more than 130 million customer events to learn a unified representation of customer behavior. Rather than building separate, task-specific models, KFM operates as a common sequence-modeling backbone that encodes behavioral patterns across different touchpoints of a customer’s interaction history with Klarna.

We will walk through the technical design of the system, including event tokenization, sequence construction, and Transformer architecture, as well as how the learned embeddings are exposed as a shared layer for multiple downstream use cases.

Attendees will gain a concrete understanding of how foundation models can be applied to behavioral sequence data in fintech, and how a single model can serve as a reusable, extensible representation layer across heterogeneous modeling tasks.

About the Speaker:
Valeria Verzi is a Principal AI Scientist at Klarna, where she leads the development of next-generation models, from foundation models to advanced deep learning or agentic systems. She has nearly a decade of experience applying machine learning to high-stakes decisioning across fintech and insurance. Before her current role, Valeria served as Klarna’s Head of Data Science for Fraud Prevention. Prior to Klarna, she held senior data science roles at Generali and NTT Data. Valeria holds a Master’s degree in Computer Science Engineering from Politecnico di Milano.

***

See you there! 🚀
The PyData Milano Team