About the role#
As a Co-Op in Reaction Prediction and Benchmarking, you will work with the Molecular Intelligence team to develop and evaluate machine learning methods for small molecule design and synthesis. You will pursue a focused research project in either reaction prediction and benchmarking or molecular property prediction for medicinal chemistry optimization. Your specific project will be determined by your interests, experience, and current team priorities.
What you'll need#
- Current enrollment in a graduate degree program (MSc or PhD) in Chemistry, Computer Science, Physics, or a similar field.
- Proficiency in Python for data processing, model training, and evaluation.
- Familiarity with cheminformatics libraries such as RDKit.
- Understanding of basic machine learning concepts, including supervised learning and model evaluation metrics.
- Experience with chemistry concepts through coursework or research.
- Experience with molecular property prediction, multi-task learning, self-supervised pre-training, or model fine-tuning is a plus.
- Familiarity with reaction data formats, reaction SMARTS, or reaction databases like Pistachio, Reaxys, or SciFinder.
- A background or interest in organic or medicinal chemistry.
What you'll do#
- Build benchmarks to assess reaction prediction models and reaction plausibility.
- Develop and fine-tune molecular property predictors using project-specific data.
- Produce reproducible data and evaluation workflows.
- Document benchmark results and research recommendations to inform future model development.
- Plan and execute your Co-Op project in collaboration with the team.
Location & details#
- This is a full-time, paid Co-Op position.
- The role is based on-site in Cambridge, Massachusetts or San Francisco, California.
- The term is rolling.


