Writing

Insights

Technical essays on machine learning engineering, AI systems, and translating models into business outcomes.

·10 min

How Predictive Modeling Changed Marketing Investment Decisions at Enterprise Scale

The gap between a model that predicts well and a model that changes decisions is where most ML projects fail. Lessons from building investment optimization systems at AB InBev.

Marketing ScienceOptimization
·8 min

Monotone Constraints in XGBoost: When Business Logic Must Override the Data

In demand estimation, higher prices must produce lower demand. No exceptions. How monotone constraints enforce economic priors in gradient boosting models and why they matter for production systems.

XGBoostDemand EstimationML Engineering
·12 min

Building ML Pipelines on Azure and Databricks: Practical Lessons

What I learned deploying ML systems on Azure and Databricks at AB InBev — from feature engineering at scale with PySpark to reproducible experiments with MLflow.

AzureDatabricksMLOps