About the Role
We are pioneering tabular foundation models, aiming to do for structured data what LLMs have done for language. This role is foundational data science, focused on building the core of tabular foundation models to enable a single model to solve diverse data science problems. The work involves inventing new frontier tools and building the dataset and benchmark bedrock.
Responsibilities
- Invent and build frontier tools to extend TabPFN, enhancing its thinking, scaling, and agentic capabilities.
- Develop new methods for a single model to generalize across various data science problems.
- Set research direction by identifying valuable model capabilities and benchmarks to pursue.
- Incorporate external research and customer needs to shape new model and tooling directions.
- Publish frontier results to advance the field.
- Build trustworthy benchmarks using structured data from real-world, high-impact problems.
- Implement baseline and competitor models to establish a standard for applied data science.
- Create an automated, agentic pipeline with human-in-the-loop oversight to scale data and benchmark foundations rigorously.
Requirements
- Proven ability to solve data science problems across diverse domains and datasets to a high standard.
- Proficiency in using the ML toolbox, including gradient-boosted trees (e.g., XGBoost) and deep learning.
- Understanding of common dataset defects (e.g., leakage, label noise, distribution shift) and their impact.
- Enthusiasm for foundational work, valuing dataset and benchmark bedrock alongside frontier tooling.
- Experience tackling challenging problems that others have overlooked.
- Thrives as a senior individual contributor in an ambiguous, early-stage, low-process environment.
- Strong judgment and opinions on best practices in Data Science for complex problems.
Skills
- Tabular foundation models
- Machine learning
- Gradient-boosted trees
- Deep learning
- Dataset defect analysis
- Benchmark development
- Agentic pipelines
- Human-in-the-loop systems
Location
- Berlin
- Freiburg
- New York
- Remote (exceptional cases)
Work Type
- Onsite
- Remote
Experience Level
- Senior
About the Company
- We are pioneering tabular foundation models, transforming structured data with AI.
- Our TabPFN v2 model is a Nature cover story and sets the state of the art in tabular machine learning.
- We have achieved significant adoption across research and industry, with applications in healthcare, finance, and infrastructure.
- We are scaling tabular foundation models to handle large datasets and demanding production environments.
- Our team comprises experienced engineers, researchers, and GTM specialists from top tech companies and research institutions.
- We are led by renowned AI researchers and advised by world-leading experts.
- We raised €9m pre-seed funding led by Balderton Capital.
- We foster a culture of rapid iteration, rigorous research, and high standards.
- We believe in the importance of in-person collaboration but are open to remote work in exceptional cases, with frequent company offsites.
- We are committed to building diverse teams and creating an inclusive environment.
Equal Opportunity
- We welcome applications from people of all identities and walks of life, especially those who may not check every box.
- We are committed to creating a safe, inclusive environment and providing equal opportunities regardless of gender, sexual orientation, origin, disability, or any other personal trait.
