
We're training the next generation of protein foundation models.
We’re a seed-stage team in San Francisco building robotic systems to automate real-world lab workflows. Most lab processes today are still manual, slow, and error-prone.
Our goal is to build a fully autonomous, “black-box” lab — systems that can take in high-level inputs and run physical workflows end-to-end without human intervention.
We’re not there yet. The system is still evolving, and a big part of this role is figuring out what actually works and making it reliable.
Anthrogen develops protein foundation models to develop new-to-nature peptides with applications from human health to frontier manufacturing efforts. They are comprised of scientists from Columbia, Duke, Berkeley, and more.
They have recently trained a large, multimodal foundation model that is SOTA across a variety of benchmarks and can even be prompted with natural language. They continue to run scaling experiments and are working towards a public launch of their first-generation models in the near future.
The team is seed-stage and has raised $4M from investors like Paul Graham, Box Group, Y Combinator, Wayfinder, and others. www.anthrogen.com