Ryan Solava
Data Scientist

Former Professor of Mathematics and Computer Science, turned Data Scientist. Solving problems with data powered solutions, and putting machine learning to use.

Skills
Programming Languages
Python
SQL
Java
C++
Haskell
Perl
Libraries
Pandas
Numpy
sklearn
GeoPandas
Beautiful Soup
Selenium
Streamlit
Keras
spaCy
Vader
NLTK
Natural Language Processing
Stemming/Lemmatization
TFIDF
Topic modeling
Sentiment analysis
Supervised Learning
Classification
Linear regression
Tree-based methods
Ensemble methods
Unsupervised Learning
K-means
K-nearest neighbors
Recommendation systems
Other
MongoDB
Tableau
Network Analysis
Git
Education
Vanderbilt University
Ph.D. Mathematics
University of Notre Dame
B.S. Computer Science & Mathematics 2012
Experience
Metis
Remote
Data Scientist
Sept. 2021 to Current

Completed a 28-week intensive data science and machine learning bootcamp. Topics included exploratory data analysis, web scraping, regression, classification, natural language processing, deep learning, and data engineering. Selected projects include:


A Board Game Recommendation Application

  • Created a full data pipeline for a board game recommendation
  • Scraped data for over 100,000 games using Selenium and Beautiful Soup
  • Stored and accessed the data using SQL
  • Built an interactive web application with Streamlit to recommend games using a custom similarity measure


Filtering Toxic Comments on Social Media

  • Preprocessed text documents by cleaning, stemming, and creating a document term matrix (TFIDF)
  • Performed topic modelling and dimensionality reduction, and fed the results into a tree-based classifier
  • Iteratively improved upon these results with a variety deep learning models which involved feed forward neural networks, RNNs, and transfer learning all using keras


Forecasting Hit and Run Accidents in Chicago

  • Performed data analysis and feature engineering on data from multiple sources
  • Visualized findings with matplotlib and Seaborn, and GeoPandas for geospacial data
  • Trained and compared a variety of classification models including XGBoost and other tree-based models using sklearn


Can Anyone Find Me a Nice Latte?

  • Processed coffee shop reviews with tokenization, lemming, and CountVectorizer
  • Compared a several dimensionality reduction algorithms (LDA, LSA, and NMF)
  • Used cosine similarity of topic vectors to generate recommendations
  • Built a web application through Streamlit to display recommendations dynamically.


Where Have our Players Gone? Predicting User Churn and Retention from chess.com

  • Processed the data with Python, Pandas, and Excel
  • Developed and pitched a proposal that considered financial and other benefits to the company,  limitations, and potential pitfalls
  • Displayed the results with interactive visuals through Tableau



Saint Mary's College
Notre Dame, IN
Assistant Professor of Mathematics and Computer Science
Aug. 2019 to Current

  • Taught a variety of courses in both math and computer science including an introduction to programming (in Python), data structures (in Java), calculus (including multivariate calculus), and graph theory. 
  • Advised multiple projects with undergraduate students each year. Topics included: decision trees, recommender systems, graph theory
  • Spearheaded assessment reform throughout the college, streamlining how data was collected to facilitate data based improvements to the curriculum


Vanderbilt University
Nashville, TN
Graduate Student
Aug. 2014 to Aug. 2019

  • Taught two courses and served as teaching assistant many times.
  • Researched, wrote, and defended my dissertation, details below.


On the Fine Structure of Graphs Avoiding Certain Complete Bipartite Minor

  • Wrote Python programs to generate all graphs with a given property up to some size (about 1,000,000 graphs generated), and to verify their structure.
  • Built on known results to develop a mathematical proof, a rigorous chain of reasoning, demonstrating the desired result
  • Communicated these technical results to people with a variety of backgrounds through both writing and presentation