Yves Ketemwabi Shamavu

Own end-to-end development of a new Azure control-plane primitive enabling rack-scale GPU networking isolation over InfiniBand and NVLink for AI training clusters
Designed Azure API for GPU platform evolution: rack-by-rack scaling, deterministic node placement, and capacity reservation across 5,000+ GPU server clusters
Technical advisor and primary interface for Tier-0 customers (OpenAI, Microsoft AI, Black Forest Labs)
Drove root-cause investigation of production InfiniBand fabric issues using data-driven hypothesis testing
Responsible AI Champion: Led red teaming across 20+ Microsoft products, establishing safety standards for AI agents impacting billions of users
Mentored PMs on problem framing, hypothesis-driven execution, and technical decision-making

Identified unmet customer needs through telemetry, competitive analysis, and stakeholder engagements
Translated insights into prioritized roadmaps driving measurable improvements in product engagement and satisfaction
Partnered with engineering and design to ship features across the Azure compute stack

Developed recurrent neural networks (RNNs) and dynamic Bayesian models for clinical decision support
Filed patent on medical AI application; published peer-reviewed research in biomedical informatics
Engineered ML training pipelines on HPC clusters (SLURM, MPI, CUDA)
Dissertation: trustworthy, clinical-grade AI

Applied advanced analytics to clinical data; designed and built full-stack EHR systems
Published research on health record system enhancement

Led multi-site humanitarian program operations across multiple countries
Built M&E frameworks and database systems (C#, SQL) improving data accuracy and program outcomes

Sub-contracted for Vodacom on network/data engineering
Built applications (Java, C#) and managed internet services for university communities