Michael Pabon

I lead a team of Data Scientists who focus on driving personalization, optimization, and intelligence across our media and marketing ecosystem to elevate both customer experience and business performance. - Develop machine learning models for audience targeting, recommendation, and marketing optimization. - Partner with marketing, product, and data engineering stakeholders to translate business needs into scalable data products that improve engagement, retention, and monetization. -Architect and oversee experimentation frameworks for AB testing, causal inference, and uplift modeling to measure true marketing impact. -Develop AI solutions — leveraging LLMs, embeddings, to create agents that optimize our workflows or serve as stand alone marketing products. -Champion data science best practices, MLOps standards, and reproducible workflows across distributed computing environments (AWS, PySpark, Databricks). -Mentor a growing team, fostering a culture of curiosity, technical rigor, and creative problem-solving. Core Technologies & Tools: Python · PySpark · AWS Glue · Databricks · Snowflake · SQL · scikit-learn · LLMs

Fandango, Rotten Tomatoes, NBC Sports Next Personalization & Recommendation Engines: - Designed and deployed a scalable, serverless ML pipeline leveraging AWS Glue and PySpark FM classifiers to deliver user/movie recommendations. - Scaled to process ~110 million users weekly, predicting up to 14 billion watch probabilities across 130 movies. - Enhanced model accuracy by 2.23x (PR AUC) and reduced runtime by 12x-38x (depending on movie slate), achieving operational efficiencies and reduced cloud costs - Streamlined the codebase, reducing lines of code by 44% compared to the legacy system. - Led creation of Golf course recommendation engine, personalizing marketing channels for millions of users based on collaborative filtering - Personalization of engagement marketing campaigns by tailoring offerings and creative based on predicted churn - Created customer centric data store which computes +50 customer behavioral metrics daily for use in marketing and product personalization Content Similarity Models: - Built advanced models utilizing movie metadata (e.g., genres, actors, MPAA ratings) to generate movie similarity scores. - Enabled recommendation engines and user-facing “More like this” features, driving higher customer engagement. Management: - Executed the key projects highlighted above, either personally or through delegation. - Manage a multidisciplinary team comprising Data Scientists, contract Data Engineers, an - ML Cloud Architect, and Data Science Interns. Report directly to the VP of Data Strategy and Monetization, ensuring alignment with organizational objectives. - Establish and monitor annual and quarterly goals to drive team success. - Cultivate and strengthen stakeholder relationships across departments. - Deliver presentations on machine learning initiatives and performance metrics to diverse audiences, including C-suite executives. - Utilize Agile methodologies and Jira for effective project management and task allocation.

GolfNow, SportsEngine, Fandango Build B2B and B2C products and services (internal and external) using techniques from machine learning statistics, econometrics and industrial engineering related to marketing and pricing of digital assets under the NBC sports umbrella.

This program had a strong focus on applying data science to the economic domain. The course work revolved around data science, statistical inference, machine learning, profit-optimization and equilibrium, consumer behavior, real-analysis, and economics.