How Tamarind Bio is Building the Infrastructure for AI-Powered Drug Discovery

Tamarind Bio, developing state-of-the-art molecular AI models usable out of the box, with no ML or infrastructure expertise required, so scientists can focus on drug discovery and research, led by Deniz Kavi and Sherry Liu, has announced $13.6M fundraise, including a $12M series A led by Dimension Capital, with participation from Y Combinator.

As AI becomes embedded in real drug discovery workflows, no longer confined to experimental projects, life sciences organizations need more than another model. They need infrastructure that enables every scientist to access models securely, efficiently, and at scale.

Demand has been strong from the outset. What began as a solution for a single Stanford lab has quickly evolved into a trusted platform for molecular AI inference, now supporting 8 of the top 20 pharma companies, hundreds of biotech and academic institutions, and tens of thousands of scientists.

Over the past year, Tamarind has grown 700%, welcoming leading organizations including Bayer, Boehringer Ingelheim, Adimab, Mammoth Biosciences, and Flagship Pioneering to build on the platform.

“With AI now central to drug discovery, Tamarind was founded to let scientists design drug candidates without worrying about cloud or inference infrastructure,” said Deniz Kavi, co-founder and CEO of Tamarind Bio. “The goal is to free teams from repetitive model optimization work so they can focus on discovering novel biology.”

“AI is becoming embedded in core life science workflows, with computational tools increasingly transforming the wet lab,” said Nan Li, Co-Founder and Managing Partner at Dimension Capital. “We’re excited to support Tamarind as they build the leading ML orchestration and inference cloud for life science.”

What began as an inference product for models like AlphaFold has evolved into a full-stack platform supporting proprietary enterprise models, custom training, and multi-stage scalable pipelines.

The model library has expanded to hundreds, spanning not only open-source tools, but also internal protocols, proprietary models trained on customer data, and complex pipelines that orchestrate multiple models together.

The commitment now is to double down on building the core AI and data infrastructure powering the next generation of medicines. Open models will continue to be supported and enhanced with proprietary data, with broad, secure access prioritized for scientists everywhere.