Research Scientist, AI

Your impact

Develop generative protein foundation models that expand what scientists can design and test for difficult therapeutic targets.

As a Research Scientist, AI, you will develop new modeling approaches, train them on large-scale structural and sequence data, and determine whether they produce useful molecular designs. Your work will connect foundational AI research with the experimental evidence needed to improve it.

What you will do

• Develop generative models for protein design. Explore and implement architectures and learning objectives that connect molecular sequence, atomic structure, and biological function.
• Design rigorous research experiments. Form hypotheses, select meaningful baselines and metrics, run controlled studies, and investigate why methods succeed or fail.
• Train and scale models. Improve training stability, efficiency, and performance as models, datasets, and computational demands grow.
• Build reliable research systems. Write maintainable research code, improve shared training and evaluation tools, and make important results reproducible.
• Evaluate model performance and limitations. Assess the quality of designed proteins, test generalization, and investigate when and why models fail.
• Close the loop with protein design. Work with computational and experimental scientists to select candidates, interpret results, and turn experimental evidence into better models and research questions.

What you bring

You are a rigorous and hands-on researcher who can move between mathematical ideas, working implementations, and empirical evidence. You pursue important questions independently while sharing context and building with the scientists and engineers around you.

• A Ph.D., or an M.Sc. with at least two years of relevant research experience, in Computer Science, Mathematics, Physics, Computational Biology, or a related quantitative field.
• Experience developing and training deep learning models, with strong foundations in probability, linear algebra, and optimization.
• Experience with generative modeling, particularly diffusion or flow matching.
• Strong Python skills, experience implementing models in PyTorch, and the ability to write maintainable research code.
• A record of research contributions through publications, open-source work, or industry projects.
• Scientific rigor, initiative, and the ability to communicate and collaborate across disciplines.
• Authorization to work in the United States.

Prior experience in biology is not required. We value researchers who can learn the scientific context, challenge their own assumptions, and follow evidence beyond the boundaries of their original discipline.

Experience that would help

• Protein modeling, structural biology, antibody design, or biological datasets.
• GPU optimization with Triton or CUDA.

Our working environment

In our small team, research scientists work across ideas, implementation, scaling, and evaluation rather than handing work between narrowly defined roles. You will collaborate with computational biologists, experimental scientists, and software engineers to move from a research question to a trained model and then to molecules the team can evaluate. We are an international team with hybrid working arrangements in Culver City.

Why Pythia

You will have the freedom to pursue foundational AI questions and the opportunity to see how the resulting models behave in real protein design campaigns. Experimental feedback makes the impact and limitations of the research concrete.

We are a small, interdisciplinary team. We value people who take ownership, move between disciplines, check assumptions against evidence, and bring others into decisions. You will have room to shape both the work and how the team works.