FOUNDATION MODELS FOR BIOMANUFACTURING
Terraferma builds specialized AI foundation models for protein production.
Our models and strains generalize to unseen proteins, achieving state-of-the-art titers across multiple protein classes.
Have a protein you need made? Contact us
Our ultra-high-throughput genetic screens collect millions to billions of reproducible datapoints per experiment. Terraferma generates and maintains one of the largest proprietary datasets linking genotype to expression phenotype.
Trained on this data, our models predict and generate high-performing sequences that work even for proteins they’ve never seen. Our self-improving pipeline means our models become more powerful with every experiment we run.
Our model-designed strains deliver up to >10× titer gains on the first try and >100× over off-the-shelf strains, reaching double-digit titers for even difficult to produce proteins.
On shelves since 2024, Ten Lives features the fastest precision-fermented animal protein to reach consumers, backed by a completed safety study. Cat nutrition is one of the most cost-sensitive markets, and Terraferma’s high-titer strains enabled Ten Lives to offer sustainable animal protein at cat food prices for the first time.
Since 2026, Terraferma has expanded its platform to benefit AI protein design, food, enzyme and therapeutic partners across the protein economy.
We work with AI protein design, food, enzyme and therapeutic teams on targets that don’t express well enough to be worth making. Bring us the target; we’ll find the strain that makes it worth producing.
Proteins that never expressed well enough to be worth producing. We design strains that get them to viable titers, expanding your candidate pool to proteins you’d otherwise have shelved.
Already in production? We raise titers on existing targets, cutting your cost per kilogram without changing your protein.
We test your protein target in our top-performing strains and report measured titers against a baseline. We can also send cell-free supernatant for your own testing.
We run custom screens for your target, using our in-house pipeline.
License our strains, or co-develop with us and receive protein product at commercial scale.
Our MIT and Harvard scientists, published in Nature and Molecular Cell, are based in San Francisco. Our dozens of expert partners in fermentation and biomanufacturing span across the United States.
Victoria Mo, MS
Victoria has over a decade of experience ranging from finetuning SOTA machine learning models to shipping both digital and physical products.
Victoria received her MS in Computer Science - Machine Learning from Columbia University.
Dr. Ruby Yu
Ruby has spent the past decade investigating gene expression in organisms ranging from yeast to humans, and she is an industry-leading expert on high-throughput genetic screens for machine learning.
She received both her BA and her PhD in Biological and Biomedical Sciences from Harvard University, and completed her postdoc at Gladstone Institutes and UCSF (Marson Lab).
Dr. Emma Kowal
Emma has over a decade of experience in molecular and computational biology. She specializes in high-throughput assays measuring gene expression, end-to-end from library design through sequencing data analysis.
She received her PhD in Biology from MIT (Burge Lab) and BA in Chemical and Physical Biology from Harvard University (Church Lab).
