We are seeking a motivated postdoctoral researcher to apply machine learning algorithms to analysis of existing and development of new synthetic chemistries of interest to the Navy. The research will initially focus on data availability in commercial reaction databases and other sources, design of custom machine representations for relevant chemical transformations, implementation of scripts for data retrieval and formatting, and data visualization for validation by human experts. Subsequent research may focus on strategies for training data augmentation such as on-the-fly generation of no-go reactions, neural networks for generating single-step retrosynthetic transformations, and neural networks for feasibility ranking of machine- or human-proposed synthetic pathways. Our ultimate goal is a combination of the above with sequential decision-making algorithms to design multistep retrosynthetic pathways for chemical compounds of interest to the Navy.  

 

This work will require prior experience with high-level programming languages (Python, MATLAB, etc.), familiarity with UNIX/Linux environment and tools, and basic understanding of synthetic chemistry principles and terminology. In-depth knowledge of machine learning algorithms is highly desirable but not required. 

key words

Retrosynthesis; Computer-assisted synthesis planning; Computed-aided synthesis; Reaxys; SciFinder; machine learning; Deep learning; Neural networks; Tree search; Tensorflow; Keras; Scikit-learn; Rdkit; Python; Theoretical chemistry

Citizenship:  Open to U.S. citizens and permanent residents

Level:  Open to Postdoctoral applicants