Research Scientist, Foundational Data Science

About the role

Join Prior Labs to build the foundations of tabular foundation models that can solve data‑science problems across the board. The role splits between inventing frontier tools for TabPFN and constructing the dataset and benchmark bedrock they rely on.

Key responsibilities

• Invent and build frontier tools that extend TabPFN, enhancing its scaling, thinking, and agentic capabilities.
• Define research direction by selecting model capabilities and benchmarks worth pursuing.
• Integrate external research and real‑world customer needs, publishing results that advance the field.
• Create trustworthy benchmarks from high‑impact structured‑data problems.
• Implement baseline and competitor models to gauge TabPFN’s performance and improvement areas.
• Develop an automated, agentic pipeline with a human‑in‑the‑loop to scale data and benchmark creation.

Required profile

• Proven track record solving data‑science problems across many domains, optimizing performance over a suite of tasks.
• Comfortable using the full ML toolbox, achieving strong results with gradient‑boosted trees (e.g., XGBoost) and deep learning.
• Deep understanding of dataset defects such as leakage, label noise, distribution shift, duplication, and mislabeled targets.
• Passion for foundational work, valuing datasets and benchmarks as much as tooling.
• Thrives as a senior individual contributor in an ambiguous, early‑stage, low‑process environment.

Required skills

• XGBoost
• Deep learning
• TabPFN