Machine Learning Scientist – Generative Molecular Design

Role overview Develop generative machine learning methods that help researchers design molecules against specified property goals. The role covers the full path from exploring chemical space and integrating evaluation signals to validating data and delivering usable research capabilities, in collaboration with computational scientists, machine learning researchers, engineers, and discovery teams.

Responsibilities - Architect and extend a generative platform for molecular design and discovery. - Build and optimize approaches such as diffusion, flow matching, GFlowNets, or language model and agent-based methods to search chemical space. - Integrate learned property predictors and simulation-derived evaluations into generation, including reward designs that account for competing objectives. - Develop computational workflows for generating and evaluating hypotheses, and communicate results that inform project direction. - Find and prepare training and evaluation data, taking responsibility for ingestion, quality control, and validation. - Work with research, discovery, engineering, and product colleagues to make models reliable and practical for scientific use.

Requirements - Four or more years of experience building and delivering generative or applied machine learning for molecular or materials discovery, in industry or equivalent research. - MSc or PhD in computational chemistry, physics, computer science, or a related or equivalent discipline. - Practical experience developing scalable machine learning with PyTorch, TensorFlow, JAX, or similar frameworks. - Ability to write robust, maintainable production code and contribute through collaborative software engineering practices. - Familiarity with cloud compute, data processing, or storage platforms. - Strong communication and collaboration skills in a multidisciplinary, fast-moving environment.

Nice to have - Active learning or Bayesian optimization for iterative design with expensive simulations or experimental assays. - Guided molecular generation, including constraints for feasible chemical space, inference-time steering, or model adaptation through fine-tuning or reinforcement learning. - Agentic or language model approaches to discovery.

Benefits and work setup The listed package includes competitive salary, equity, performance incentives, health, dental, and vision coverage, retirement plan matching, parental leave, flexible hybrid work, and paid time off. Mentorship and dedicated learning budgets are also described.