AI Biologist - Longevity

About the Role

You will be the scientific authority on ground truth, testing whether AI agents can reason like an expert aging biologist. Today's agents struggle to separate biological age from chronological age, spot confounds that make aging signals look real, and judge which longevity claims hold up. Working with real published and public datasets, you'll build the tasks behind our benchmarks, starting with aging biomarkers and the hallmarks of aging. Your judgment sets the standard agents are held to.

Build the task

• Turn real aging research questions into tasks, and write gold-standard answers an expert would defend
• Curate data across molecular and population scales: DNA methylation, transcriptomics, proteomics, single-cell, longitudinal cohorts and biobanks
• Build tasks around aging clocks, biological age prediction, senescence, epigenetic drift, and mitochondrial dysfunction
• Partner with software engineers to turn your judgment into workflows

Set the hard calls

• Test whether agents catch cell-type composition, batch effects, and age imbalance between groups
• Test whether agents know when findings in worms, mice, or cell models won't translate to human aging
• Adjudicate contested longevity claims: what the data support, what they don't, and what would settle it

Judge the agent

• Challenge reasoning that sounds right but wouldn't survive peer review or a go/no-go decision
• Pinpoint where agents fail, what an expert would do instead, and what evidence would change the call
• Reward calibrated uncertainty over confident guesses

Requirements

• Master's or PhD in biology, bioinformatics, computational biology, or a related field
• Research in aging biology, or in a field where aging is studied: cancer biology, epigenetics, immunology, metabolism, mitochondrial biology, neurodegeneration, or epidemiology
• Working knowledge of aging: the hallmarks, how clocks are built, and biological vs. chronological age
• Depth in at least one of: Computational: machine learning or statistics on omics data; Experimental: hands-on experience generating omics or aging data (sequencing, model organisms, senescence or mitochondrial assays), so you know how experiments go wrong
• You code your own analysis end to end in Python or R
• Work you drove yourself (first-author paper, thesis, independent project) that you can defend in detail
• A track record of catching errors in data that others missed
• Daily use of AI coding tools, and the confidence to overrule an AI when the evidence demands it

Strongly preferred

• Published research in aging, geroscience, or longevity
• Experience building or validating aging clocks, or analyzing lifespan or healthspan data
• Familiarity with UK Biobank, TCGA, GEO, Tabula Muris Senis, or the NIA Interventions Testing Program