Giovani Moctezuma Rodríguez León

FaceMatch is a proof of concept for a face-based identification system developed as a Third Horizon Initiative within the university technology office at Arizona State University. I developed and set up all the back-end code and infrastructure, using Python and AWS tools.

A public package and open-source Python software development kit for a data labeling company to seamlessly integrate with their back-end infrastructure. I contributed to the development of new features, testing, and documentation while following coding best practices.

A full site for job search focused on IT. As the machine learning engineer, I designed and developed an automated matching algorithm that combines on-site algorithms with third-party tools like IBM Watson to enhance performance. The matching feature was a big differentiator against competitors.

The redesign and implementation of the matching algorithm produced results two orders of magnitude faster, from two minutes to less than 10 seconds per prediction. The biggest challenge was joining data from different sources and third-party APIs to produce high-quality predictions.

An offline, machine learning-powered app to automate entrances and exits at sports centers. I designed and developed the whole solution, using CNN-based algorithms for face recognition on the back end with a WinForms UI as the front end.

The software helped to manage sports center partners and eliminate losses due to pending payments or expired memberships. The solution was capable of recognizing people with 99% accuracy (based on public datasets).

An in-depth study of consumer trends for a retail store. As the primary data scientist, I performed customer segmentation techniques using clustering and data visualization to extract powerful insights. This involved analyzing millions of records to identify trends based on customer demographics and customer behavior such as buying patterns.

As a principal architect and developer, I built a FastAPI back end powering SportFX, an AI-driven sports performance platform. The system integrates AWS SageMaker for real-time video analysis, delivering biomechanical insights, 3D overlays, and LLM-generated coaching feedback.

I designed a multi-tenant architecture with Auth0 authentication and role-based access. The video pipeline processes uploads through ML models, extracting performance metrics and recommending personalized drills. Features include Stripe subscription management, gamification with achievements and streaks, notification systems, and affiliate tracking.

The production system emphasizes reliability with structured logging, health monitoring, and comprehensive testing. Containerized deployment on AWS ECS with CI/CD pipelines supports thousands of analyses with sub-second response times.