Menu
User documentation
Publications|Stories|2026

Alps Empowers Scientists: Imanol Schlag

They design AI systems that make the most of our world-class supercomputer. They see the Alps as a platform for collaboration and a foundation for sovereign, independent AI. They use it as a mathematical microscope to explore the complexity of molecular interactions and to make quantum effects visible. They are our users. In this series, we introduce some of the people behind these projects and explore how they are using Alps to advance science and benefit society.

September 3, 2026 - by CSCS

Name
Imanol Schlag   

Position
AI Research Scientist at ETH AI Center, co-lead of the Apertus project, and lecturer at ETH Zurich.

Area of research
Artificial Intelligence 

My focus
My research focuses on building large, open, and responsibly developed AI language models, and on improving the fundamental architectures that power them.

“Our work directly contributes to technological sovereignty” - Imanol Schlag (Image: Alessandro Della Bella, CSCS)

What supercomputing and Alps mean for me
Alps gives us a unique opportunity to ensure that frontier AI development is not the exclusive domain of a handful of private companies, but something that public institutions can meaningfully contribute to. For me, this is about the democratisation of the defining technology of our time. 

What challenges do I face
Building a frontier language model is as much an engineering challenge as a research one; it requires the coordinated effort of dozens of highly trained experts across data, infrastructure, compliance, and training. Much of the critical work, such as stabilising training runs on thousands of GPUs, is essential but does not produce publications, which creates a tension with academic incentive structures. Bridging the gap between research and engineering cultures, each with different goals and reward systems, is one of the most underappreciated challenges in academic AI development at scale. 

What I like most about my work
What excites me most is that our work directly contributes to the technological sovereignty of the public, ensuring that a technology as consequential as AI is not solely controlled by a few private entities, but remains accessible and transparent for society at large. At the same time, I find it deeply motivating to witness how AI is beginning to transform virtually every domain of science, from healthcare to climate research to the humanities. Being at the intersection of building this technology and shaping how it is developed responsibly is a rare and privileged position. 

My favourite recent project using Alps
Apertus is a family of multilingual language models (with 8 billion and 70 billion parameters, respectively) trained on 15 trillion tokens in over 1,000 languages. What sets it apart is full openness: we release the weights, the training code, and complete data documentation — all under a permissive Apache 2.0 license. The models are designed to comply with Swiss law and the EU AI Act, including respect for opt-out mechanisms and prevention of data memorisation. Apertus provides a trustworthy foundation for research and industry to build AI applications without relying on opaque proprietary systems. 

Career background
Dr. Imanol Schlag is an AI Research Scientist at the ETH AI Center, co-leading Apertus, the Swiss AI Initiative's LLM effort. He studied at FHNW and the University of St Andrews before completing his PhD at USI/IDSIA under Jürgen Schmidhuber (2023). His research focuses on open-source LLMs, neural architecture innovations (particularly fast weight programmers like the DeltaNet), and responsible AI development. He has conducted research at Meta FAIR, Google Research, and Microsoft Research, and currently teaches Large-Scale AI Engineering at ETH Zurich.