Predictive Modeling of "How Much" (Regression)
Business Examples: Forecasting how many units, dollars, or any other measure you care about is going to occur at a point, or multiple points in the future. Optimized pricing for a product or service given the customer, the job, and other factors based on our previous deals.
Biological Examples: Predicting catalytic abilities of enzymes, PK/PD, predicting metabolite yield, and immunogenic distance of antigens.
Chemistry Examples: Predicting continuous properties of chemicals/polymers/materials such as solubility, melting point, conductivity, tensile strength and more.
Predictive Modeling of "Good or Bad" (Classification)
Business Examples: Will a borrower default on a loan, will a customer churn, will a deal fall through, will someone respond to direct marketing?
Biological Examples: What genes are predictive of a disease, what is the function of protein, is a given protein a good target for a vaccine, is a residue part of an epitope, will a protein be highly digested during isolation?
Chemistry Examples: Will a small molecule be toxic, have a certain packing structure, inhibit a protein, elicit a desired biological response, have a desired taste, make a strong polymer, or exhibit other useful properties?
Accelerating Tasks with Generative AI
Business Examples: Fine tuning models to give accurate and informed answers to accelerate solutions engineering, competitive intelligence, customer support, and more.
Research Examples: Fine tuning LLM's on research literature and proprietary data to suggest next steps, troubleshooting advice, research gaps, applications, and more.
Business Example: What other items would a customer want based on previous actions
Biological Example: Based on virus surveillance sequencing what virus co-occur or precede other viruses
In addition to the biological use cases of machine learning mentioned above, I also have experience in the more traditional assembly, alignment, differential expression, and clustering of RNA and protein expression data.
Hosting predictive models and automated process through R-Shiny apps or Python Streamlit apps.