Machine learning, quantitative modeling, and data engineering.
Machine learning, quantitative modeling, and data engineering.
Vincent Applied Sciences is the consulting practice of Mike Vincent, Ph.D. I build predictive and quantitative models, test how well existing ones perform, and construct the data systems that both depend on. The work spans machine learning, statistical and Bayesian methods, and mathematical modeling, and the right approach depends on the problem rather than on a preferred toolkit.
I build predictive models and machine learning systems, Bayesian and classical statistical models, and mathematical or simulation models of a process. This includes feature design, model selection, and the engineering required to run a model on real data rather than only in a notebook.
I measure how well an existing model performs against real outcomes. This work covers calibration, uncertainty quantification, out-of-sample testing, and backtesting against historical data. Evaluations are designed in advance so that the result is informative whether or not a problem is found.
I design and build reproducible pipelines and analytical databases. Each system rebuilds from its original source files rather than being edited in place, and includes validation checks that identify problems in the incoming data before those problems reach an analysis.
I deliver written work that explains what was found, what methods were used to test it, and how much confidence the evidence supports. Where a result is too uncertain to act on, the report says so directly.
I hold a Ph.D. from Northwestern University and have spent my career in biomedical engineering and computational research. I have published more than thirty peer-reviewed papers and hold several patents, one of which has been licensed commercially. Before founding Vincent Applied Sciences, I established and led the research and development group at Grove Biopharma and worked as a scientist at the Van Andel Institute.
My work has spanned biotechnology and the life sciences, including nanotechnology and synthetic chemistry, as well as equity markets and company evaluation, sports and prediction markets, and consumer software. I also design and operate production software, which means a model I build can be deployed and maintained rather than handed over as a prototype.
Please include a short description of the problem you are working on and the timeline you have in mind.