Principal AI Scientist; f​/m​/d Mixture-of-Architecture Foundation Models

Staff/Principal AI Scientist

At SAP Business AI, we are exploring the next frontier of AI on structured business data like graphs, tables and more. Foundation Models on structured data, like SAP-RPT-1, have ushered in the GenAI era for business relevant scenarios. Now, we are pushing further. We want to enable companies across all industries to collaboratively train foundation models and build knowledge graphs. As Staff/Principal AI Scientist (f/m/d), you will play a leading role in defining, prototyping and iterating a composable, multi-modal architecture for foundation models spanning across structured (tables, time series, graphs,...)

and unstructured data (text, code, images, …). You stay up to date with the most recent research in the highly dynamic realm of foundation models and you use these insights to work out ideas for technical concepts, model architectures and inform backlogs and research roadmaps of the model R&D teams. You can easily switch between deep hands-on work, training and evaluating foundation models, and conceptual work, defining technical visions and roadmaps for scaled execution by multiple R&D teams.

This is the ideal role for a highly focused, ambitious individual with both, a strong technical and coordinating abilities to grow in a dynamic, fast-paced environment, and your chance to become a key contributor to one of the very few leading foundation models of European origin.

Define research roadmap for mixture-of-architecture foundation models to enable training across multiple modalities of business data, such as tables, time series, graphs, text and others

Define processes, methods, hypotheses, and experiments with the ambition to lead the project together with a team of AI Scientists

Balance long-term research roadmap with and hands-on coding, experimentation, and technical improvements to bootstrap the development

Guide the development of enablement materials and blueprints for training mixture-of-architecture models on different, potentially federated data sources

Collaborate cross-functionally to inform value proposition and partnering ambitions

PhD or Master's degree in Computer Science, Artificial Intelligence, or other relevant disciplines

Candidate must have a strong background in Foundation Models and/or World Models, with a substantial publication record, including recent top-tier conference papers, or equivalent experience delivering models in an applied research environment.

Experience with foundation models for large structured data or multimodal foundation models is a strong plus.

Demonstrated hands-on experience in researching, prototyping, and advancing machine learning models

Proficiency in Python, and experience with ML frameworks such as PyTorch, Tensor Flow, or similar.

Optionally, hands-on experience with Graph ML frameworks

Experience with Machine Learning and Knowledge Modeling on structured data such as in the ERP or CRM domain is a plus

Proven history of exemplary leadership and strategic mindset, complemented by superior organizational abilities

Strong communication and documentation skills for technical and non-technical audiences – you can explain technical model architecture in detail just as well as you can explain the why and the benefits for users and business stakeholders