Our focus is at the nexus of deep learning, physics, and chemistry. We are architecting a new class of foundation models that understand the language of molecules, pushing the boundaries of generative AI to make these previously impossible medicines a reality.
These models power our own pipeline of drug candidates and accelerate our partners’ most challenging discovery programs.
Our Core Research Areas
Our research is organized around solving fundamental problems in machine learning that are critical for drug discovery. We pioneer novel deep learning architectures and methods, with active research and development in areas including:
Genesis has developed a leading foundation model for 3D biomolecular structure prediction, called Pearl. We leverage proprietary architecture, equivariant neural nets and novel modeling techniques to achieve unprecedented precision.
We go beyond natural language, developing bespoke LLMs that understand the grammar of molecular structure and chemical reactions to generate novel, optimized molecules.
We use advanced reinforcement learning techniques to steer our generative models, allowing them to solve complex, multi-objective optimization problems that are intractable with other methods.
We build massive, multi-modal foundation models that integrate diverse data sources—from 3D structures to bioassay results—to predict the complex ADME properties essential for a successful drug.
Our research advances state-of-the-art equivariant networks that inherently understand the physics of molecular systems, ensuring our models respect rotational and translational symmetries.