Founding Machine Learning Scientist

About Tabula

Tabula is building an AI-first therapeutics company.

We are starting with bacteriophages, natural predators of bacteria, and building the models and experimental systems needed to design better therapies for hard-to-treat infections. The immediate problem matters on its own. Antibiotic resistance is a large and growing global problem, and new approaches are badly needed. But we also think this is the beginning of something broader.

Most biotech companies are mostly biologists with a small computational team attached. We think that model is going to look dated. Our view is that drug and therapy discovery will become much more computational over time, and we are building Tabula around that belief from day one.

Two things make this a particularly interesting problem for machine learning. First, we own our data generation loop. We are not just consuming static datasets and hoping they are good enough. We can generate new data, learn from it, and improve the system over time. Second, phages are one of the rare places in biology where the path from model output to human impact can be unusually short.

We picked this area in part because, relative to most of biotech, the feedback loop to real-world use is fast.

One way to describe what we are doing is simple: we are using machine learning to help design living therapies that can save lives. That sounds a little like science fiction, which is part of why we think it is worth doing. But it is also a very practical engineering and research problem.

We are hiring a Founding ML Scientist to help build the machine learning core of the company.

This is a research-oriented ML role. We are looking for someone who is strong in modern machine learning, likes difficult technical problems, and wants to work on something where the connection between model quality and real-world impact is unusually direct.

You will work on model development, experimental design, evaluation, and iteration in close partnership with our wet lab. That collaboration is central to how we work. The lab exists in large part to generate and validate the data that improves our models. Over time, we expect that loop between model design, data generation, and biological validation to become one of the company’s core advantages.

We also picked this problem deliberately. A lot of biologically important ML work sits very far from actual deployment in humans. In many cases, even very strong model progress may take years to affect a patient. Phages are different. They give us a much shorter path from model output to something that can matter in the clinic. In bio time, that is unusually fast.

Because this is a founding role, the job is not just to run experiments inside an existing system. The job is also to help define what the system should be. You will help shape how we think about research direction, model quality, experimentation standards, and the relationship between the computational and biological sides of the company.

• Design and run model experiments against difficult biological data

• Help define the research roadmap for how Tabula’s models should improve over time

• Work closely with the lab to shape what data gets generated and why

• Build better practices around experimentation, evaluation, and reproducibility

• Contribute to technical decisions around model design, training, and research direction

• Help a small team build an AI company whose output is not content or software alone, but real therapies for real patients

• Strong background in modern machine learning and deep learning

• Fluency in Python and PyTorch, plus comfort with the broader modern ML stack

• Ability to reason clearly about model architectures, training, evaluation, and tradeoffs

• Good research judgment about which ideas are worth testing and how to evaluate them

• Comfort working in an early-stage environment where the problems are hard and the path is still being defined

• Ability to work closely with domain experts outside ML

• Strong interest in applying machine learning to a problem in the physical world, not just at the software layer

We do not require prior biology experience. In fact, we do…