Summary
Ada Rosa is a software engineer with 13 years of experience building large-scale AI and cloud systems, currently driving runtime and infrastructure work for running LLMs on AMD hardware at Microsoft Azure. She led the first end-to-end GPT-4 run on AMD, authored a high-performance GPTQ kernel with 4x throughput gains, and designed reproducible containerized AI experiment tooling that now powers millions of core hours and tens of millions of downloads. Her work spans low-level performance (HIP/ROCm, MLIR), orchestration (Kubernetes, Go), and research-inspired algorithms that parse 400TB/month of logs to detect complex failures and improve SLAs. She has a history of cost-saving systems engineering—e.g., $3M/month savings through caching and storage optimization—and has shipped reliability improvements used by 60M daily users. Beyond production systems, she explores bio-inspired modeling of personalized human behavior and has background in real-time GPU algorithms for VR and robotics, reflecting a blend of research curiosity and production impact.
13 years of coding experience
5 years of employment as a software developer
Bioengineering and Biomedical Engineering, Bioengineering and Biomedical Engineering at Self Schooled
University of Illinois Urbana-Champaign
Illinois Mathematics and Science Academy
English