(Senior) Machine Learning Scientist,Materials Composition/Structure

Full-time - Entry Level

Cambridge, MA

About the position

The position at Flagship Labs 97 Inc. (FL97) involves pioneering the application of artificial intelligence in materials discovery and development. The role focuses on training, fine-tuning, and deploying deep learning models to connect materials composition, structure, and performance, contributing to a sustainable economy. The candidate will work in a collaborative, cross-functional environment, driving innovation in scientific methods through AI and data-driven approaches.

Responsibilities

• Train, fine-tune and deploy deep learning models connecting materials composition, structure and performance.
• Develop and train deep learning architectures for representation learning and generative AI over materials composition and structure.
• Develop physics-informed learning architectures and loss functions that capture conservation laws and other invariances.
• Connect information retrieval with LLM tools and quantitative mathematical and physical symbolic reasoning.
• Develop and implement strategies to optimize machine learning models for materials synthesis and performance prediction.
• Utilize AI-backed methods for lab orchestration, experimental assay design, and optimization of process parameters in materials synthesis and testing.
• Contribute to a digital platform that continually fine-tunes models as more data becomes available, driving constant improvement.
• Work closely with experimental teams to drive material discovery and development.
• Communicate findings to stakeholders through written reports, slide decks and verbal presentations.

Requirements

• Strong proficiency with PyTorch, including training and deploying models, preferably with experience in multi-GPU parallelization.
• Demonstrated expertise in training supervised or unsupervised deep learning models, preferably on structure or composition of materials or chemicals—such as in crystals, polymers, or biomolecules.
• Expertise in including physics-based inductive bias in deep learning model architecture or loss function: conservation laws, symmetry (equivariance), functional form, PINNs, neural ODEs.
• Proven track record of publishing scientific papers in scientific journals or ML conferences, or contributing to/creating public code bases related to machine learning and materials science.
• Proficiency in Python and the data science ecosystem (NumPy, SciPy, Pandas) and data visualization (matplotlib, plotly, etc).
• PhD in Computer Science, Applied Mathematics, or a quantitative discipline, with a strong focus on machine learning.
• Excellent communication skills for conveying technical findings to diverse audiences.

Nice-to-haves

• Experience with cloud computing services (e.g., AWS) to optimize training and evaluation processes.
• Familiarity with integrating machine learning in experimental workflows within materials science or chemistry.