Job description

WHO WE ARE Foundation models transformed text and images. Structured data - the largest and most consequential data format in the world - stayed untouched, until now. What LLMs did for language, we're doing for tables. We pioneered tabular foundation models: TabPFN v2 was a Nature nature.com cover story, has passed 3.5M+ downloads and 7,500+ GitHub stars, and runs in production from detecting lung disease with Oxford Cancer Analytics oxcan.org to preventing train failures with Hitachi siliconangle.com. The hardest problems - millions of rows, real-time inference, entirely new modalities - are still open, and no one else is working on them at this level. We're a small, highly selective team of 40+ priorlabs.ai with backgrounds from Google, DeepMind, Meta, Apple, Amazon, Jane Street, and CERN, led by Frank Hutter linkedin.com, Noah Hollmann linkedin.com, and Sauraj Gambhir linkedin.com, and advised by Bernhard Schölkopf and Turing Award winner Yann LeCun. In July 2026, less than 18 months after our €9M pre-seed, we joined SAP priorlabs.ai as an independent frontier AI lab - same team, mission, and open-weights models, now backed by more than €1 billion over four years. What you'll do This role is foundational data science: building the foundations of tabular foundation models so a single model can solve data-science problems across the board. Roughly half the work is inventing new frontier tools for TFMs, and half is building the dataset and benchmark bedrock they stand on. - Invent and build the frontier tools that extend TabPFN, including its thinking, scaling, and agentic capabilities, and the new methods that let one model generalize across the full landscape of data-science problems. This is the most open-ended part of the work and grows over time. - Set the research direction by deciding which model capabilities and benchmarks are worth pursuing, choosing what is worth solving rather than optimizing a score someone else set. - Bring in external research and real customer needs to shape new model and tooling directions, and publish frontier results that move the field forward. - Build trustworthy benchmarks from the structured data behind real, high-impact problems, so the team optimizes for real-world performance rather than one leaderboard. - Faithfully implement the baselines and competitor models that set the gold standard of applied data science, giving the team a read on where TabPFN leads and where there is room to improve. - Build an automated, agentic pipeline with a human in the loop so this data and benchmark foundation scales to far larger volumes without losing rigor, itself a genuinely new tool. What we're looking for - You have solved data-science problems across many domains and datasets to a high standard, optimizing for strong performance across a whole suite of tasks rather than the single best score on one. - You work undogmatically across the ML toolbox, including getting strong results with gradient-boosted trees (such as XGBoost) and not only with deep learning. - You understand the common categories of dataset defects (leakage, label noise, distribution shift, duplication, mislabeled targets, and similar) and why each corrupts a training or benchmark signal. - You are energized by foundational work, valuing the dataset and benchmark bedrock as much as the frontier tooling, and you have taken on hard problems others passed over. - You thrive as a senior individual contributor in an ambiguous, early-stage, low-process environment. You are opinionated on best practice in Data Science and can make good judgement calls on approaches to complex problems. Nice to have - Experience building or extending evaluation harnesses, benchmark suites, or experiment frameworks that others rely on. - Experience building LLM- or agent-assisted pipelines with a human in the loop to scale a previously manual workflow. - Experience acting as the link between external research or customer needs and an internal model or product roadmap. - Prior work on tabular, structured-data, or foundation-model problems, or helping shape an emerging research subfield through community work. LIFE AT PRIOR LABS You'll work alongside researchers and builders who hold themselves to a very high bar - in the quality of their work and in how they work with each other. We move fast and still take the time to do things right. Our teams are based in Berlin, Freiburg, and New York - when you're working on something as hard as TabPFN, being in the same room matters. But great people come from everywhere, and in exceptional cases we're open to remote, which usually means frequent travel to one of our offices. Wherever you're based, the whole company comes together regularly for offsites to build and celebrate together. OUR COMMITMENTS The best products and teams are built by people with a wide range of perspectives and backgrounds. We welcome applications from all identities and walks of life - especially if you've ever felt discouraged by "not checking every box" - and provide equal opportunities regardless of gender, sexual orientation, origin, disability, or any other trait that makes you who you are. We care about how your data is handled - see our Recruiting Data Privacy priorlabs.ai page

Requirements

Must-haves

Set the research direction by deciding which model capabilities and benchmarks are worth pursuing, choosing what is worth solving rather than optimizing a score someone else set.
Bring in external research and real customer needs to shape new model and tooling directions, and publish frontier results that move the field forward.
Build trustworthy benchmarks from the structured data behind real, high-impact problems, so the team optimizes for real-world performance rather than one leaderboard.
Faithfully implement the baselines and competitor models that set the gold standard of applied data science, giving the team a read on where TabPFN leads and where there is room to improve.
Build an automated, agentic pipeline with a human in the loop so this data and benchmark foundation scales to far larger volumes without losing rigor, itself a genuinely new tool.
You have solved data-science problems across many domains and datasets to a high standard, optimizing for strong performance across a whole suite of tasks rather than the single best score on one.
You work undogmatically across the ML toolbox, including getting strong results with gradient-boosted trees (such as XGBoost) and not only with deep learning.
You understand the common categories of dataset defects (leakage, label noise, distribution shift, duplication, mislabeled targets, and similar) and why each corrupts a training or benchmark signal.
You are energized by foundational work, valuing the dataset and benchmark bedrock as much as the frontier tooling, and you have taken on hard problems others passed over.
You'll work alongside researchers and builders who hold themselves to a very high bar - in the quality of their work and in how they work with each other. We move fast and still take the time to do things right.
We care about how your data is handled - see our Recruiting Data Privacy https://priorlabs.ai/recruiting-data-privacy page

Nice-to-haves

Experience building or extending evaluation harnesses, benchmark suites, or experiment frameworks that others rely on.
Experience building LLM- or agent-assisted pipelines with a human in the loop to scale a previously manual workflow.
Experience acting as the link between external research or customer needs and an internal model or product roadmap.
Prior work on tabular, structured-data, or foundation-model problems, or helping shape an emerging research subfield through community work.

Application prep

Use the role’s language and application details while tailoring your materials.

Keywords to mirror

aimlllm

Platform: ashby
Effort: quick