I was 1 of 3 selected to work with Merck under UC Berkeley's Data Discovery Program. I collaborated with Merck Data Engineers and Senior Scientists, leveraging data analysis techniques to generate a benchmark dataset for machine learning analysis, to be used company-wide.
Data Transformation: Converted over 1M protein sequences into machine-readable vectors using Ablang embeddings.
Dimensionality Reduction: Applied advanced techniques such as UMAP and t-SNE to visualize principal features and reveal meaningful patterns in the data.
Dataset Generation: Extracted biologically relevant information and developed a benchmark dataset for discovery biologics.