At Ubisoft, I own an agentic AI system end to end, from architecture and deployment to continuous optimization. I also build test suites to validate accuracy, safety, and reproducibility within strict security boundaries.
Previously, I owned the infrastructure for a real-time LLM toxicity-detection service on hybrid AWS and on-prem Kubernetes. I cut p99 latency from about 500 ms to under 100 ms, improved reliability, and rebuilt the cluster as code. At Université Laval, I built hybrid financial simulation infrastructure and an ML platform for research and education.

