Scientific Software Developer | Materials Informatics Specialist
Current𝐌𝐚𝐭𝐞𝐫𝐢𝐚𝐥𝐬 𝐒𝐜𝐢𝐞𝐧𝐜𝐞 𝐒𝐨𝐥𝐮𝐭𝐢𝐨𝐧 𝐆𝐫𝐨𝐮𝐩, 𝐌𝐚𝐭𝐞𝐫𝐢𝐚𝐥𝐬 𝐈𝐧𝐟𝐨𝐫𝐦𝐚𝐭𝐢𝐜𝐬 • Lead and mentor teams of junior scientists on different materials informatics and digital transformation projects• End-to-end deployment of ML models based on supervised machine learning, deep learning, computer vision, NLP, Bayesian optimization, active machine learning, generative models, and genetic algorithms to deliver custom solutions to clients. • Develop custom Python software, data pipelines, CI/CD pipelines, and web applications. • Help develop instructional materials for best practices in software development, MLOps, and other topics for internal and client use𝐏𝐫𝐨𝐣𝐞𝐜𝐭 𝐇𝐢𝐠𝐡𝐥𝐢𝐠𝐡𝐭𝐬Photochemical Design• Used a combination of graph neural networks (GNNs), generative models, and retro-synthesis models (i.e., de novo molecular design) with an interface web application to help chemists explore the chemical space for different photoactive chemicals.Surrogate Model API and Bayesian Optimization of SiC Products• Built a custom Python library to automate different high-throughput simulation workflows and implement Bayesian optimization and active machine learning loops to support SiC processing development.• Developed an interactive web application to prepare simulation pipelines, manage project data, and explore data-driven insights for chemical researchers. Computer Vision Quality Control• Integrated a model training pipeline for instance segmentation models in existing quality control (QC) workflows to segment SEM images of different chemical products and automate downstream QC calculations. • To support this model and reduce the burden of collecting the training data, a web application was developed to aid the annotation of unlabeled images using the Segment Anything Model.