Data Scientist

Mark D. Borysiak, Ph.D.

********.*@*****.***

206-***-****

Seattle, WA

Github: https://github.com/mborysiak/Portfolio

LinkedIn: https://www.linkedin.com/in/mark-borysiak-63117611/

TECHNICAL SKILLS

Neural Networks, Tensorflow

Regression and Classification

Clustering, Dimensionality Reduction

Machine Learning

Boosted, Ensemble Algorithms

SQL (Postgres, SQLite)

Statistics, Data Analytics

Python Stack (Pandas, Numpy)

PySpark (SQL, MLlib packages)

Data Wrangling

EDUCATION

University of Washington PhD, Chemical Engineering

Ohio State University BS, Chemical Engineering

PROFESSIONAL EXPERIENCE

Phoresa Co-Founder & Scientist July 2016–Aug 2017

• Developed medical device and statistical algorithm to expedite patient diagnosis and care

• Generated Bayesian economic model to identify optimal customer sub-segments based on 50+ interviews with medical directors, physicians, nurses and other relevant stakeholders

• Raised $500k in private financing commitments and awarded $225k Small Business grant (SBIR)

University of Washington Graduate Research Associate Sept 2011–Aug 2016

• Created method to accurately and conveniently classify patient disease status using experimental design and quantitative image analysis on 100+ GB of data

• Produced work results leading to 8 peer-reviewed publications and 2 US patent applications

• Wrote research grant based on thesis project that received $2M+ in funding over 4 years

DATA SCIENCE PROJECTS

Machine Learning to Predict Fantasy Football Scores

• Created a machine learning pipeline with 30% greater draft ranking accuracy compared to community consensus using web scraping, feature engineering, and ensemble modeling

• The model can be leveraged to optimize decision-making for creating and managing teams

Classifying Trip Distance for Bike Ride Share

• Utilized public datasets to predict the trip distance for Divvy ride shares using LightGBM

• Uses feature engineering and model selection to minimize computation and maintain accuracy

Time Series Forecasting with ARIMA and LSTM Neural Networks

• Compared ARIMA and LSTM Neural Network algorithms for 7 different time series

• Methods and algorithms would be useful in predicting churn, user-volume, and other metrics

Additional: Factors Associated with Teen Pregnancy, Predicting Housing Prices in Iowa

Contact this candidate