A cell crossing a microscope frame is easy to overlook. To Jacob Kimmel, its movement was a clue. During his doctorate at the University of California, San Francisco, he built ways to infer a cell’s state from time-lapse images. The question was wonderfully direct: if you watch closely enough, can behavior tell you what is happening inside? His thesis turned that question into statistical and machine-learning methods. Years later, with a much bigger team and a CEO title, he still works from a similar premise. First find a way to see clearly. Then decide what the observation permits you to do.
Kimmel is the co-founder and CEO of NewLimit, a South San Francisco biotechnology company built around epigenetic reprogramming. His career has moved through university research, open-source software, a laboratory at Calico Life Sciences, and now a company with substantial capital and a large scientific staff. These are different jobs. They also form a remarkably continuous line. He studies systems that change, builds tools to describe the change, and asks whether the description can guide an intervention. The scale keeps expanding; the intellectual habit does not.
The habit is visible in the way he presents himself. His LinkedIn summary is a single sentence about working “at the intersection of atoms and bits.” It sounds like a slogan until one reads his CV. There are papers about cell behavior and single-cell data; code repositories for classification and analysis; then a progression of roles that steadily joins computation to experiments. A founder biography often rushes past the technical apprenticeship. In this case, the apprenticeship is the plot.
The microscope was a measuring instrument
At UCSF, from 2015 to 2018, Kimmel worked with Wallace Marshall and Andrew Brack. His doctoral thesis, “Inferring stem cell state from cell behavior,” dealt with a difficulty that reaches well beyond any one experiment. A scientist can take a snapshot of a cell and measure its features, but a snapshot hides the path by which the cell arrived there. Time-lapse imaging supplies the path. Kimmel worked on methods to extract information from those moving images and connect visible behavior to a changing internal state.
There is a small philosophical wager in this work. The microscope gives you pixels, not an explanation. A model can find patterns in the pixels, but it must still earn the trust placed in it. Kimmel’s papers from that period address both sides of the problem: how to quantify motion and how to understand what the resulting numbers mean. His research also extended into cell-state transitions and the different ways cells change over time. The scientist in these papers is not simply collecting data. He is trying to make observation precise enough to carry an argument.
Even a short detour fits the pattern. In autumn 2017, he was a deep-learning research intern with IBM Research’s Cell Engineering Group in San Jose. There he developed neural-network methods for time-lapse imaging and worked on better cell tracking. It is a modest line on a crowded CV, but it connects the patient work of microscopy to the computational tools that would become central to his later career. He was learning to ask what an algorithm could add to an experiment, and what an experiment could reveal about the algorithm.
A scientist who shipped tools
After UCSF, Kimmel joined Calico in South San Francisco. He began as a data scientist, became a computational fellow, and then led a laboratory as a principal investigator. The titles mark a genuine change in responsibility. A data scientist can solve a defined problem. A lab leader has to choose which problems deserve a group’s time, give the work a direction, and make several kinds of expertise fit together. Kimmel’s own account of the lab says it combined computational and experimental approaches to study cell identity and reprogramming.
This period also produced work with a more public life than a journal article alone. Kimmel co-developed scNym, open-source software for classifying single-cell data. His GitHub profile still displays it among his pinned projects, alongside other tools with names such as velodyn and heteromotility. The list is a reminder that scientific ideas do not travel only through papers. They travel through software that another researcher can run, criticize, adapt, and improve. His scNym work received a top-paper award at a 2020 computational-biology workshop associated with the International Conference on Machine Learning.
“Developing therapeutics at the intersection of atoms and bits.”Jacob Kimmel, LinkedIn profile
The phrase has a practical meaning here. “Atoms” are the experiments and materials that refuse to behave like a clean spreadsheet. “Bits” are the models, code, and data that help people decide what to test next. Kimmel’s career has repeatedly moved between the two. He appears comfortable in the friction. A model may suggest a direction, yet a result still has to survive the laboratory. A laboratory may produce a surprising result, yet the team needs a way to learn from it systematically.
The founder’s job gets bigger
NewLimit brought Kimmel together with Brian Armstrong and Blake Byers. The company describes its founding across 2021 and 2022 in different public materials; what is clear is that Kimmel joined at the beginning and became its scientific leader. An early LinkedIn announcement introduced him as head of research. NewLimit’s current company page traces the sequence from head of research to president to CEO. That is not a routine change of stationery. It is the path of a scientist taking responsibility for the whole organization.
The public record gives glimpses of how he approaches that responsibility. NewLimit publishes regular progress updates under his name. They are dense with the mechanics of research: what the team measured, how the experimental process changed, and where models helped choose work. In a field that can attract grand forecasts, this running account is more revealing than a polished promise. It shows a leader who wants the intermediate steps on the record, not only the final announcement. His interviews likewise spend real time on the limits of biological prediction and the difficulty of turning elegant science into a repeatable process.
Money changed the scale of the task. NewLimit announced a $130 million Series B in 2025 and a $435 million Series C in June 2026. The latter round was led by Founders Fund. Those numbers make a bold headline, but money is only permission to attempt more work. It does not make the experiments easier or the organizational decisions smaller. By July 2026, Kimmel was writing about manufacturing scale, model performance across different cell types, and a team recruiting across research, development, and operations. The company was becoming a larger machine, with more opportunities for the pieces to miss one another.
One July update described a model that could draw on data from multiple cell types. In one comparison, the team said it matched the performance of a more specialized model using roughly a third as much data from the cell type under study. That is a technical result, and an early one. Its relevance to Kimmel’s story lies in the feedback loop: experiments produce data; models use the data to guide choices; the next experiments test whether the guidance was useful. The original microscope question has grown into an organization-wide method.
The person outside the diagram
A CV can make even an interesting life look like an orderly sequence of institutions. Kimmel’s personal site adds a few welcome irregularities. He says he likes writing, climbing big hills, and playing several instruments with “equitable levels of mediocrity.” The joke works because it has no interest in polishing the résumé. He also offers an email address for conversations about biology, machine learning, carbon capture, and the relative merits of San Francisco coffee shops. The last item sounds especially suitable for rigorous peer review.
His writing has a life beyond company updates. He has posted reading lists and essays on his personal blog, including books on scientific history. The choices are not a secret key to his character, and they need not be. They show someone interested in how discoveries happen over time, how institutions fund them, and how a few stubborn questions can occupy a career. Those interests sit naturally beside the open-source projects and the painstaking progress reports. He seems to prefer a record of the work to a mythology about it.
The contrast between his public-facing roles is part of the appeal of the story. On a podcast he can talk expansively about biological prediction and the economics of research. On his website he can laugh at his own musicianship. In a progress report he can move from hiring to a detailed explanation of experimental efficiency. None of these voices cancels the others. Together they make a plausible portrait of a scientist who became an executive without entirely leaving the bench-side way of thinking behind.
The next experiment is a company
Kimmel now leads a company that must coordinate scientists, engineers, operations specialists, and outside partners. Its stated plans place a first human trial in 2027. The schedule is a company target, not an outcome already achieved, and it brings an entirely new kind of scrutiny to the work. The research questions remain difficult. The management questions may be just as demanding: how to keep a growing team candid about uncertain results, how to decide which measures matter, and how to preserve speed without letting the evidence grow thin.
There is no tidy answer in the available record, and Kimmel has not written one. His own announcement of the 2026 financing ended with the line, “We are still at the beginning of the story.” It is a fair description of both the company and its CEO. The boyish charm of a founder tale would be to call the destination inevitable. The more interesting account is this one: a researcher trained himself to read change in a moving cell, built tools to make those readings useful, and now has to lead an organization whose changes are harder to fit inside any frame.
A microscope can hold a cell in view. A company needs a different kind of attention. Kimmel’s career suggests that he knows the difference, and also knows the value of the old question. Watch carefully. Measure honestly. Let the next move answer to what you actually saw.