Sovereign AI in Practice: Building, Evaluating and Scaling Multilingual LLMs

Karin Sevegnani
Senior Solutions Architect
NVIDIA

Karin Sevegnani is a senior solutions architect at NVIDIA, where she leads the NVIDIA AI Technology Centre (NVAITC) in the UK. In this role, she drives collaborations with higher education institutions and research organizations to advance AI innovation and adoption. Before joining NVIDIA, Karin worked as a Research Engineer in Edinburgh, applying her expertise in AI development. She holds a PhD in Conversational AI from a joint program between the University of Edinburgh and Heriot-Watt University, completed under a combined scholarship and degree arrangement. Her specialization lies in Natural Language Processing (NLP), particularly in conversational AI systems.

Vincent Van Steenbergen
CEO
ThinkDeep AI

Vincent has more than 20 years' experience in software engineering, data processing, and distributed ML/DL pipelines. With ThinkDeep he works on the DeepBrain platform to allow organizations to build a single source of truth and extract business insights from their unstructured data (documents, videos, and audio).

André Martins
VP of AI Research
EuroLLM

Andre Martins is an associate professor at the Computer Science Department (DEI) and the Electrical and Computer Engineering Department (DEEC) at Instituto Superior Técnico. He is also the VP of AI research at Unbabel in Lisbon, Portugal, and a senior researcher at the Instituto de Telecomunicações, where he leads the SARDINE Lab.

Jens Grivolla
Senior Research Engineer
Barcelona Supercomputing Center

Jens Grivolla is a senior research engineer at the Barcelona Supercomputing Centre. As part of the Language Technology Lab's innovation group his work centers on bridging the gap between research and real world applications. Through collaborations with public administration, companies, or the internal development of proof-of-concept prototypes, he aims to validate and demonstrate the technologies developed at BSC, including automatic translation and speech technologies or the fully open Large Language Models from the Salamandra and ALIA families, covering a large range of languages with a focus on European and Iberian languages.

Elie Bakouch
Research Engineer
Hugging Face

Elie Bakouch is a Research Engineer at Hugging Face, where he is responsible for pre-training topics. He holds a Master's in Science from Telecom SudParis. Elie joined Hugging Face in 2024. Prior to this, he completed an internship at CEA, where he worked on Federated Learning. 

Elie has made significant contributions at Hugging Face, including the state-of-the-art SmolLM 1 & 2 models, which were open-source and outperformed other models of similar size, and the fully open-source replication of Deepseek R1, known as open-r1.

Adam Henryk Grzywaczewski
Senior Deep Learning Data Scientist
NVIDIA

Adam Grzywaczewski is a senior deep learning data scientist/applied research scientist at NVIDIA, where he supports customers scaling and deploying deep learning workloads. He specializes in large-scale training—not just on deep learning systems and software design, but also on algorithms that allow for large batch data/model/pipeline parallel training.