This was to be a primarily lab-based position, but I worked in the Process Chemistry Development department, supporting the effective and economic scale up of drug manufacturing from lab to production scale. Takeda Pharmaceuticals produces and develops drugs to treat rare illnesses. While minimizing computational resources used. Performed clustering analysis and dimensionality reduction to visualize chemical space and select molecules of interest to optimize knowledge gained Predict reaction yield, built with PyTorch and scikit-learn. Data was stored and updated using NoSQL (MongoDB).ĭeveloped and productionized a supervised machine learning model using graph neural networks and random forest trees to evaluate features and

Quantum-mechanics based descriptors in a supervised machine learning model.Ĭonstructed a data pipeline to take novel reaction data from the electronic lab notebook, generate an optimized 3D geometry for all molecules,Īnd calculate various cheminformatic and quantum-mechanics based descriptors.

My primary project was developing and productionizing a tool for chemists to predict the reaction yield of Suzuki reactions using classical cheminformatic and I worked in the Computer-Aided Drug Discovery (CADD) group, supporting various departments in experimental design and analysis. Novartis Institutes for Biomedical Research (NIBR) is the early-stage R&D center for Novartis Pharmaceuticals. Novartis Institutes for BioMedical Research (Cambridge, MA)

Esh elyoum ashraf abdelbaki