Machine Learning Scientist I/II, Chemistry

Flagship Pioneering•Cambridge, MA

7d•Onsite

About The Position

Our start-up is building a novel chemical discovery platform to reshape how we develop formulations at the molecular level. By combining artificial intelligence with a cutting-edge laboratory discovery pipeline, our interdisciplinary team will chart an unexplored chemical landscape, enabling breakthroughs in drug delivery, agrichemical formulation, and safe, high-performance industrial chemicals. Our company is backed by Flagship Pioneering, a biotechnology origination company that has founded and developed over 115 scientific ventures over the past 25 years. Flagship Pioneering is dedicated to the development of an ecosystem of first-in-category life sciences companies, resulting in $20+ billion in aggregate value, 500+ issued patents, and more than 50 clinical trials for novel therapeutic agents. THE ROLE We are seeking a motivated ML Scientist to lead the predictive pipeline of a lab-in-the-loop platform for novel chemistry. You will define and implement models that reason through the physical rules that govern chemistry to predict the properties of physiochemical interactions at the molecular level. The ideal candidate is eager to take the lead on building a chemistry discovery platform from scratch and be the team’s computational visionary.

Requirements

PhD (or equivalent experience) in machine learning, computational chemistry, organic chemistry, chemical engineering, materials science, or related field
Strong experience in building, training, and validating ML models in chemistry, material science, or related domains
Knowledge of fundamentals of chemical properties is essential
Proficiency in Python-based deep-learning frameworks (e.g. torch, pytorch lightning), ML logging software (e.g. MLFLow, WandB), relevant scientific libraries for chemistry (e.g. RDKit), and hyper-parameter optimization (e.g. Optuna)
Ability to set up and maintain python compute environments
Strong code organization and reproducibility work ethics and usage of git-based workflows
Familiarity with cloud/HPC frameworks (e.g. aws, azure), containerization (e.g. docker), and batch compute
Experience handling large, noisy experimental datasets
Strong team player: ability to communicate and work hand-in-hand with chemists, lab scientists, and operational team members

Nice To Haves

Experience with DFT and molecular dynamics tools preferred
Experience in Bayesian optimization, active learning, reinforcement learning preferred
Prior work on lab-in-the-loop platforms or early-stage teams preferred
Knowledgeable in thermodynamics, statistical mechanics, and hydrogen-bonding interactions; familiarity with phase-diagram analysis preferred

Responsibilities

Develop models that predict phase behaviors and physiochemical properties of chemical mixtures across temperature, hydration, and concentration gradients
Build a generative “lab-in-the-loop” modeling pipeline , that incorporates human input of the chemical formulation scientist as well as experimental feedback and proposes the next set of experimental samples to be tested in the wet lab
Build on top of next-generation chemistry ML methods , such as chemical LLMs , reaction prediction , and electronic structure methods like DFT
Own the data flow pipeline, including definition of data standards and processing , data distribution analysis, and visualization to guide the team and inform decision-making processes