Nicholas David


BS Materials Science and Engineering, Additional Major in Applied Physics, Carnegie Mellon University (2022)



My research focuses on data-driven materials discovery though the use of machine learning, natural language processing, and bayesian statistics. In particular, I’m interested in developing predictive synthesis models using only positive training data. These models embrace survivorship bias, which plagues not only materials synthesis literature, but literature in other scientific domains as well. Since failed synthesis recipes are rarely published, these models use only successful, text-mined synthesis recipes as training data. Through these models, we aim to machine-learn the fundamental criteria for solid-state materials synthesis. By doing so, we will have also developed a unique framework for performing inference on datasets where survivorship bias takes precedence.

I am pursuing a joint PhD in Materials Science and Engineering and Scientific Computing, with the Michigan Institute for Computational Discovery and Engineering (MICDE).