I developed, deployed and refined a custom machine learning recommender system using collaborative filtering, factorization matrices and neural networks that led to a 26% improvement in conversion rates. I oversaw and configured the MLOps infrastructure implementation from scratch using Synapse Analytics, Machine Learning, Data Bricks, and Microsoft Azure Kubernetes to support 20 requests per second. I supervise and develop an interoperable artificial intelligence system composed of predictive models, neural networks and intelligent agents for a collection service; its implementation finally improved the weekly collection rate by 200%, which is equivalent to increasing the weekly collection from 200,000 dollars to 600,000 dollars.