Daniel Koch Email & Phone Number

Los Angeles, CA, US

• Created and maintained several key production models for a major US telecommunications account, predicting propensity to churn across multiple product specific call center queues, generating measurable incremental.
• Collaborated with data subject experts and data science teams to improve mutual understanding of client data and maximize behavioral model alignment with real-time call routing algorithms
• Applied and optimized open-source tools to score large scale models on millions of customers
• Owned and designed entire model life cycle; from pulling disparate data into a unified customer view, deriving potential predictors of behavior, training/testing/validating models with various algorithms/hyper.

• In coordination with product managers in B2B sales: collected data on customers across enterprise data warehouse, created and fine-tuned classification models, streamlined scoring process for monthly score production.
• Ultimately implemented entire end-to-end data modeling workflow from scratch, coming into a finance team that traditionally worked on reporting.

• Developed dozens of propensity models at multiple levels of customer granularity, from multiple channels. Models scored behaviors on up to 100 mil consumers and 20 mil B2B lines using 1000’s of attributes, with scores.
• Oversaw the development lifecycle of each model, from preliminary consultation with business partners, to development of sql definitions of customer behavior to query within the enterprise data warehouse, to parameter.
• Created ad-hoc data reports and analyses for different business units, requiring research in enterprise data warehouse, critical thinking in how to convert a business request into a statistical report and vice-versa.
• Formulated visualizations on customer behavior over time to help executive partners better understand the state of the business.
• Researched and tested new variables for use in predictive modeling, sourced from third-party vendors Bluekai/Axciom, requiring text extraction, variable creation, and experimentation.
• Trained/mentored interns and new hires on internal procedures and best practices in modeling workflows.