About

Approach to Data Science

My name is Zak, a self-proclaimed data science nerd who thrives at the intersection of data science and business strategy. I believe data science is a tool that can help companies operate more efficiently, make better decisions, and uncover unique insights, thanks to my diverse experiences across various functional areas.

Based in San Diego, CA, I'm a coffee-loving, golf-playing, early-to-bed kind of guy who thrives outside my comfort zone. My love for numbers extends beyond the desk—check out my Instagram for ratings of burgers, salads, and lobster rolls.

Though I've turned 30, I'm still passionate about learning. I'm pursuing a Master's degree in Computer Science and Information Technology from the University of Pennsylvania to formalize my knowledge and fill gaps in my self-taught data science skillset. Fight on, Pennsylvania!

Experience

Manager/ Senior Manager, Data Science, Corvus, San Diego, CA

• Managing a high-performing team of 4-8 data scientists at a hyper-growth insurtech startup, fostering a collaborative environment that encouraged career development and skill growth for team members.
• Developing, implementing, and maintaining a strategic roadmap in collaboration with product and engineering stakeholders to maximize speed-to-delivery and ensure alignment with company objectives.
• Leading the design, development, and deployment of multiple machine learning models focused on underwriting workflow and pricing optimization, including:
  • A random forest queue management model that provides the front-line sales team with a priority ranking of submissions, ensuring they focus time and effort on submissions with the highest likelihood of becoming sales for Corvus. High-priority cases are 400% more likely to sell than low-priority cases.
  • A Bayesian price optimization tool that leverages perimeter scanning technology to recommend discounts to low-risk insurance applicants and improve our competitive position. This model has led to a 20% increase in our bind rate (how often a submission becomes a policy on Corvus' book).
  • Currently developing a clustering algorithm to profile renewal business and carve out portions of our renewal book that are eligible for an automatic renewal, enabling underwriters to focus their attention on problematic renewals and reduce the time they spend on trivial ones.
  • Expanded the feature set in our existing proprietary cyber risk scoring model ("Corvus Score"), improving Corvus' profitability by around 4%.
• Regularly communicating model mechanics, quantifying business impact, and ensuring smooth cloud-based (AWS) deployments to a wide variety of unique audiences.
• Contributing to the technical vision of the Data Science team, implementing an experimentation framework in Jira, centralizing shared data science tasks as repositories in Github, and driving innovation throughout the organization.

Data Scientist, John Hancock Financial , Boston, MA 

• Analyzed and interpreted a large electronic health records dataset, identifying key factors affecting mortality risk for insurance applicants.
• Developed and implemented machine learning models to predict mortality risk, streamlining the underwriting process and reducing the time from days to hours.
• Presented data insights and model findings to stakeholders in a clear, concise, and actionable manner, contributing to data-driven decision-making across the organization.
• Continuously explored new tools, techniques, and best practices to enhance the efficiency and effectiveness of data analysis and model development, such as:
  • Integrating SHAP (SHapley Additive exPlanations) for better model interpretability and explainability.
  • Adopting advanced ensemble methods, including Random Forests, Gradient Boosting Machines (GBM), and XGBoost, to enhance predictive performance and stability.
  • Streamlining data preprocessing and feature engineering pipelines using tools like scikit-learn's Pipeline and ColumnTransformer to ensure reproducibility and maintainability.

Skills

I'm a self-taught data scientist who began my career in finance, where I honed my skills working with Excel spreadsheets, Word documents, and PowerPoint presentations. Over time, I have gained experience in identifying, developing, and testing hypotheses, as well as building and deploying a wide range of machine learning models, from simple to complex. While I am comfortable with model deployment, I typically leave the cloud deployment to machine learning engineers. Here's a comprehensive list of skills I've mastered:

• Machine Learning Frameworks: Scikit-learn, LightGBM, CatBoost, H2O, PyTorch
• Natural Language Processing (NLP): NLTK, spaCy, Gensim, BERT, OpenAI GPT, Transformer models
• Data Manipulation: Pandas, NumPy, Dask, Spark
• Deep Learning: Neural Networks, TensorFlow, Keras
• Programming: Java, Python, SQL, MATLAB, and familiarity with C/C++, React
• Cloud Computing: AWS, Azure
• Data Visualization: Matplotlib, Seaborn, Plotly, ggplot, Tableau, Power BI
• Version Control: Git, GitHub, GitLab, Bitbucket
• Project Management: Agile methodologies, Scrum, Kanban, Jira, Trello
• Soft Skills: Problem-solving, critical thinking, creativity, attention to detail, adaptability, communication, collaboration, presentation