AI Biologist - Microscopy

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 Biologist to join our Microscopy team, pushing what AI can achieve in biological imaging.

About the Role. You will be the scientific authority on ground truth, testing whether agents can reason like an expert microscopist. AI agents today struggle to interrogate raw images, separate real biology from artifacts in acquisition, staining, and segmentation, and reach conclusions an expert would stand behind. Experts themselves disagree on roughly 1 in 6 images, so knowing where the evidence ends is part of the work. You will work with software engineers and scientists to build the tasks and datasets behind our benchmarks across cell biology, tissue, and disease. Your judgment will set the standard agent outputs are held to, including:

Build the task

Define what expert reasoning looks like for an imaging question, and write gold-standard answers with the rationale an experienced microscopist or pathologist would defend

Design realistic scenarios across the imaging workflow: sample prep, staining, acquisition, preprocessing, segmentation, quantification, and interpretation

Curate image datasets across modalities: immunofluorescence, confocal and live-cell, electron microscopy, histology and histopathology, and high-content screening

Build analysis pipelines in Python, including deep-learning segmentation, that produce verifiable ground truth

Partner with software engineers to turn your judgment into new workflows

Set the hard calls

Test whether agents distinguish true signal from artifact, such as bleed-through, uneven illumination, fixation damage, out-of-focus planes, or segmentation failure

Adjudicate ambiguous images: decide what the data supports, where reasonable experts would disagree, and which additional stain, image, or experiment would settle the call

Test whether agents choose the right imaging modality, resolution, and unit of analysis for the question

Judge the agent

Interrogate agent reasoning step by step, and challenge conclusions that sound right but would not survive pathologist review or lab meeting

Pinpoint where agents go wrong: 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 calls on ambiguous images

Requirements

PhD in cell biology, bioimage analysis, pathology, or a related field; or an MD with pathology training

Senior, hands-on imaging experience: you have looked at thousands of samples and know what normal and abnormal look like

Direct experience with image segmentation tools such as CellProfiler, Cellpose, StarDist, QuPath, ilastik, napari, or Fiji

Deep expertise in at least one area (bioimage analysis, cell biology, or histopathology) and working fluency across the others

Strong command of experimental design, image quantification, and statistics

AI-ready from day one: daily use of Claude, Cursor, or similar agents, with the judgment to spot 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

Bonus: electron microscopy, pathology, or pharmacology experience (e.g., pharmacodynamic or drug-uptake imaging assays)

Bonus: familiarity with public resources such as the Image Data Resource (IDR), BioImage Archive, Human Protein Atlas, Cell Painting Gallery, or TCGA whole-slide images

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.