European AI Research Laboratory

EUWL

At the frontier of artificial intelligence, we pursue research that redefines what machines can understand, reason about, and create.

Who we are

We don't follow.
We found.

EUWL is a collective of researchers, engineers, and theorists united by a single conviction: that artificial intelligence should be developed openly, rigorously, and with deep respect for its societal impact.

We operate across five European cities, combining the continent's strongest traditions in mathematics, philosophy, and computer science to build AI systems that are not just powerful — but principled.

340+

Researchers

6

Years active

5

Labs

Core Capabilities

What we excel at

Autonomous Reasoning

Systems that decompose complex problems, formulate multi-step plans, and execute decisions with minimal human oversight.

Multimodal Understanding

Unified models that process and connect text, images, audio, and structured data within a single coherent representation.

Distributed Learning

Federated and distributed training paradigms that scale across geographies while preserving data privacy and sovereignty.

Safety Alignment

Rigorous alignment methods ensuring AI systems remain interpretable, controllable, and aligned with human values at every scale.

By the numbers

Scale of ambition

0

Parameters

0

Researchers

0

Publications

0

Patents

Active Projects

Research stack

01

Project Helios

A multi-modal foundation model trained on scientific literature, code, and experimental data. Helios can reason about novel hypotheses and design experimental protocols for verification.

multimodal science reasoning
02

Project Nyx

Privacy-preserving federated learning at European scale. Nyx enables collaborative model training across institutions without centralizing sensitive data, respecting GDPR by design.

federated privacy gdpr
03

Project Aether

Next-generation alignment research combining constitutional AI with formal verification. Aether aims to provide mathematical guarantees about model behavior in safety-critical domains.

alignment verification safety
04

Project Kronos

Temporal reasoning and long-horizon planning. Kronos develops architectures capable of understanding cause and effect across extended timeframes for autonomous decision-making.

temporal planning agents

Publications

Recent work

Emergent Reasoning in Sparse Transformer Architectures

M. Volkmann, K. Iosifidis, A. Reinholt, T. Okafor

NeurIPS 2024

Quantum-Enhanced Gradient Descent for Non-Convex Optimization

L. Petrova, S. Balachandran, J. Hämäläinen

ICML 2024

Federated Pre-Training with Differential Privacy Guarantees

D. Osei, R. Castellano, F. Lindqvist, P. Nkemelu

ICLR 2024

Safe Alignment Through Constrained Constitutional Self-Play

A. Reinholt, D. Osei, R. Castellano

AAAI 2024

Join the frontier.

We're hiring researchers, engineers, and policy specialists across all five EUWL laboratories.