Research Scientist, Virtual Cell Modelling & Perturbative Biology Foundation Models

About Valence Labs Valence Labs is Recursion’s frontier AI research engine. We lead high-impact research programs designed to materially expand Recursion’s ability to discover and develop medicines for complex diseases. Our team balances near-term pragmatism with a long-term view of where the field is heading in the next 3–5 years, incubating, designing, and productizing the approaches we believe will define the future of drug discovery. Our work is driven by optimism, purpose, and a shared vision for a healthier tomorrow. We publish in top journals and conferences, contribute to open science, and engage with some of the world’s most active ML-for-drug-discovery research communities. Our teams are based in London and Montreal, with deep ties to Mila, the world’s largest deep-learning research institute. About The Role You will be joining a research program building multimodal foundation models to predict cellular responses to chemical and genetic perturbations across petabyte-scale omics and imaging data. The work spans generative and distributional modeling, representation learning for molecules and genes/proteins, and the design of biologically grounded evaluation frameworks. The goal is to close critical gaps in the pre-clinical pipeline, replacing or augmenting wet-lab perturbation screens with in silico predictions that are reliable enough to drive drug discovery decisions. We are seeking a Research Scientist with strong ML research and engineering skills, and genuine curiosity for biology, to join a multidisciplinary team of ML researchers, engineers, and computational biologists working toward a shared goal: building virtual cells that transform how medicines are discovered. Key Responsibilities Generative Modeling: Research and develop generative and distributional models (e.g., flow matching, diffusion models) to predict high-dimensional cellular responses. Scalable ML Engineering: Build and maintain ML systems capable of processing massive multiomics datas