Machine Learning Scientist, Generative Molecular Design (Material Discovery)

Role overview Build and productionize generative machine learning capabilities that help researchers identify candidate materials matching complex property goals. The work spans model design, reward signals, data quality, and deployment, with close collaboration across machine learning, computational science, discovery, engineering, and product teams.

Responsibilities - Develop and improve generative approaches for exploring chemical space, such as diffusion, flow matching, GFlowNets, or language model and agent methods. - Connect generation systems to learned property predictors and simulation-based evaluations, designing reward signals for competing goals such as performance, safety, and environmental impact. - Create computational workflows to generate and assess material hypotheses, and use findings to guide research decisions. - Identify training and evaluation data, then manage ingestion, quality checks, and validation. - Turn research models into robust tools that discovery scientists can use, while collaborating with research and product partners.

Requirements - At least four years of hands-on experience delivering generative or applied machine learning for material discovery, in industry or an equivalent research setting. - MSc or PhD in computational chemistry, physics, computer science, or a related or equivalent field. - Experience building scalable machine learning systems with a framework such as PyTorch, TensorFlow, or JAX. - Strong production software skills, including writing maintainable code and working with modern collaborative engineering practices. - Familiarity with cloud platforms for compute, data processing, or storage. - Clear communication and effective collaboration in a fast-paced, multidisciplinary setting.

Nice to have - Active learning or Bayesian optimization for design cycles involving costly simulations or experimental assays. - Guided generation for materials, including constrained generation, inference-time steering, or domain adaptation through fine-tuning or reinforcement learning. - Experience applying agents or language models to materials discovery.

Benefits and work setup Competitive salary, equity, and performance incentives are described, along with health, dental, and vision coverage, retirement matching, parental leave, flexible hybrid work, and paid time off. Professional development offerings include mentorship and learning budgets.