Senior Data Scientist ✨ Machine Learning 🚀 Model Development 🔧 Tool Development 💻 Software Programmer 📦 R Software 📈 Leadership

"Making predictions is easy ... making accurate ones is much more difficult." -- Me --

I love to solve problems.

Often the problem can be understanding a complex biological process, but it can also be as simple as fixing something that's broken (e.g. a door that jams, a bicycle, or even machine learning software). In particular, I like to apply my data science skills to better understand, or even solve, the problems we face.

Over the past 12+ years I have combined my statistical knowledge and Open-Source Software tools to solve complex problems in the Life Sciences proteomics (high dimensional) space. In so doing, I have created a comprehensive R-based machine learning analysis ecosystem that standardizes and enables biomarker discovery and predictive model development.

Sometimes the problem is inconsistency across teams or analysts ... thus I promote adherence of "tidy" data principles and am a strong proponent reproducible research and use of bioinformatics pipelines.

Other times the problem can be sharing results across the organization ... thus developing an Application Program Interface (API) infrastructure that enables anyone to access model results with ease.

With my teaching background, I find it important to mentor junior team members while simultaneously leading more senior members. This collaborative spirit is essential to building and effective team that delivers to stakeholders, fosters a sense of accomplishment, and drives revenue generation.

I am always open to discuss possible roles 🔭 and whether my skill set can solve problems in your space. Please reach out via:

How	Where
📫
☎️	720.259.9982
🔗	www.linkedin.com/in/stu-field-sr-data-sci

Machine Learning 🚀	Statistics 📊	Open-Source 💻	Software Tools 🔧
Random Forest	Logistic regression	R	Linux🐧, MacOS 🍎
Naive Bayes	Linear regression	C++	Git, GitHub
Lasso/ridge regression	GLMMs	Python 🐍	AWS
k-Nearest neighbour	Mixed-effects models	LaTeX	BASH, GNU
PCA	Survival analysis	CI/CD	BitBucket
Ensemble methods	Multivariate statistics	Docker 🐋	Slack
Maximum Likelihood	ANOVA		Kubernetes

• Data Analysis: created high-dimensional, high-throughput, multi-plex, proteomics machine learning analysis ecosystem which enabled (and standardized) biomarker discovery and model development across analysts.
• Project Leadership: led highly successful Open-Source Software (OSS) initiative enabling customers to not only understand highly complex analysis concepts in the proteomics space, but to conduct those analyses themselves.
• Analysis Reports: generated standardized analysis templates enabling reproducible research and results across the organization.
• Leadership: successfully led a team of 3-5 direct reports through analyses, code review, self-enablement, and career development.
• Written Accomplishment: proven ability to summarize complex analyses via strong publication record.

Topic 🚀	Thumbnail 📈
False Discovery
Mixture Models
Logistic Regression
Naive Bayes
The Birthday Paradox
Mack-Wolfe Tests
Mixed Effects
Monty Hall Paradox
Decision Boundaries
Class Imbalance

Topic 🚀	Thumbnail 📈
Pitch Classifier

• 💬 Favorite food: 🐟 🌮
• 📚 I am currently learning woodworking🪵 ... I'm not very good, but I can make a lot of sawdust!
• 💬 Ask me about: bikes and R ... I'll talk your👂 off!
• 🚴 I'm an avid cyclist: come say hi on

• I maintain several R software libraries (📦) that implement statistical and machine learning techniques in biomarker discovery. Some of my popular published CRAN 📦 are:
  • SomaDataIO
  • SomaPlotr
  • gitr
  • helpr
• These projects support analyses in the general health care (Life Sciences) space to generate proteomic based clinical insights in health spaces such as:
  • cardiovascular disease
  • liver disease (NASH/NAFLD)
  • alcohol effects
  • biological aging
  • exercise status
  • metabolic disease
• Favorite techniques:
  • logistic regression (ol' faithful)
  • random forest
  • naive Bayes
  • KKNN (nearest neighbor)
  • survival analyses
  • ensemble methods
• I am a proponent of the open-source software, conducting the majority of my research/analysis via Linux toolkits, R, and the RStudio IDE.
• I promote conforming to the adherence of so-called "tidy" data, a philosophy of data science designed to share underlying data structure, grammar, and format which facilitates the generation of reproducible analyses.

• https://github.com/MartinHeinz/MartinHeinz
• https://dplyr.tidyverse.org/articles/programming.html