Meet Hugo

Portfolio

⚡ Industry-Leading Inference Performance

• Delivered 865,000 tokens/sec and 1.1M tokens/s of real-world inference throughput with respectively the NVIDIA NVL72 GB200 GPUs and the NVIDIA NVL72 GB300 GPUs Azure software stack - highlighted by Satya Nadella (CEO, Microsoft) and Jensen Huang (CEO, NVIDIA) at Microsoft Build 2025 and Microsoft Ignite 2025.
• Achieved top-ranking MLPerf Inference submissions across multiple hardware architectures (GB200, H200, H100, A100, A10).

🧠 Record-Breaking AI Training at Scale

• Led the first supercomputer-scale (10k+ H100 GPUs) submission in MLPerf Training demonstrating end-to-end scaling of transformer models - highlighted by Satya Nadella (CEO, Microsoft) at Microsoft Ignite 2023.
• Reduced the cost performance of AI LLM Training by 28% with NVIDIA H200 GPUs - highlighted by Satya Nadella (CEO, Microsoft) at Microsoft Ignite 2024.

🌐 Pioneering AI Developements in the Cloud

• Executed the first-ever public proof of concept for training Large Language Models (GPT-3, 530B parameters) at scale in the cloud using NVIDIA NeMo framework - catalyzing the global rush for AI infrastructure, including OpenAI's move to Azure - highlighted by Satya Nadella (CEO, Microsoft) at Microsoft Build 2022.
• Proved on-par performance between cloud and on-premises infrastructures across 10 foundation models — <5% variance at scale.

🔍 Industry Benchmarking Standards

• Spearheaded the Azure AI Benchmarking Guide, a widely-referenced validation suite setting performance baselines of infrastructures (NVIDIA and AMD hardware) for AI production workloads.

Current team members

• Sarah Choe - Technical Program Manager, Microsoft AI
• Elaine Fu - Technical Program Manager, Microsoft AI
• Ashley Trunko - Technical Program Manager, Microsoft AI

Latest content