AI Neuroscientist

Latch builds rigorous scientific benchmarks (benchmarks.bio) for AI agents across biology. We work with frontier labs and pharma to measure whether agents can handle real biological workflows: the messy, multi-layer reasoning that matters in the field.

We’re seeking an AI Neuroscientist to join our Neuroscience team, pushing what AI can achieve in neuroscience.

About the Role. You will be the scientific authority on ground truth, testing whether agents can reason like an expert neuroscientist. AI agents today struggle to interrogate raw evidence across molecular, cellular, circuit, tissue, and behavioral levels, separate real effects from artifacts and confounds, and reach conclusions and R&D decisions an expert would stand behind. You will work with software engineers and neuroscientists to build the tasks and datasets behind our benchmarks across normal nervous-system function and disease; your judgment will set the standard agent outputs are held to, including:

Build the task

Define what expert reasoning looks like for a research question or R&D decision, and write gold-standard answers with the rationale an experienced neuroscientist would defend

Design realistic research scenarios across the workflow, from hypothesis and target selection through model-system choice, experimental design, data processing, and inference

Curate multi-modal data bundles (electrophysiology, calcium imaging, EEG/sEEG, MRI/PET, single-cell and spatial transcriptomics, histopathology, behavior, biofluid markers)

Pair basic-science and disease tasks (e.g., hippocampal replay in health and its disruption in epilepsy or Alzheimer’s disease) to test whether an understanding of normal function transfers to pathology

Partner with software engineers to turn your judgment into new workflows

Set the hard calls

Test whether agents choose the right model system and unit of analysis, and recognize when rodent, iPSC, or organoid findings will not translate to human disease

Adjudicate contested claims in the literature: decide what the data supports, what they do not, and which experiment would settle the question

Judge the agent

Interrogate agent reasoning step by step, and challenge conclusions that are fluent and textbook-correct but would not survive lab meeting, peer review, or a go/no-go decision

Pinpoint where agents go wrong: state why a step fails, what an expert would do instead, and what evidence would change the call

Score agent confidence, rewarding explicit uncertainty over overconfident binary calls when evidence is incomplete

Requirements

PhD or MD/PhD in neuroscience or a closely related field

Hands-on neuroscience research experience with first-hand data analysis and a publication record

Direct experience analyzing raw data in at least two modalities: electrophysiology (patch-clamp, MEA, Neuropixels), optical imaging, EEG/sEEG, neuroimaging, single-cell or spatial omics, neuropathology, or quantitative behavior

Deep expertise in at least one area (systems/circuit, cellular/molecular, clinical/translational, or computational neuroscience) and working fluency across system levels

Strong command of experimental design and statistics

Proficiency in AI agent use, with the judgment to identify where model reasoning is technically correct but scientifically unsound

Confidence to overrule an AI agent when the evidence demands it, and the openness to update when shown wrong

Ability to calibrate confidence and communicate uncertainty rather than defaulting to binary calls

Practical understanding of model systems (iPSC neurons, organoids, rodents, NHP, human cohorts)

Familiarity with core resources is a plus: Allen Brain Cell Atlas/BICAN, DANDI/NWB, OpenNeuro/BIDS, IBL, ADNI, PPMI, UK Biobank, ClinVar/gnomAD

Culture @ Latch

We practice radical transparency: you’ll know how the company is doing, what’s working, and what isn’t. It’s always us against the problem, never each other. We’re opinionated but respectful of each other’s time, and we think through possible solutions before bringing a problem to anyone. We use AI constantly, so we reward original expert judgment, not AI output passed off as your own. You’ll start on the bench that fits your expertise and can move across teams as you build a track record.