Gabriel Aguilar Perez

At AMD, I build Python-based reliability forecasting software for product lines, uncovering a systematic failure pattern across 600k+ units and driving fixes projected to save $40M+ over product lifetime.

I improved flagship GPU reliability by approximately 2.1x by working with Fellow-level architects on a scalable fix adopted by two Fortune 50 clients. I also built a C++ simulation engine and system-level reliability calculator that reduced success-estimate uncertainty from 5–10% to 1–2% for multi-GPU systems.

I automate reliability analysis with Kubernetes pipelines, cutting turnaround from about 2.5 days to four hours, and built an internal HTML/JavaScript portal to streamline team workflows. I’ve also authored executive-facing customer quality reports with quantified uncertainty and recommendations used in VP-level decision meetings.

My mathematics background includes research at the University of Chicago in differential and information geometry for neural networks, alongside experience teaching and grading advanced mathematics courses.