I'm a senior principal software engineer in RedHat. I've extensive experience in data storage and performance engineering, with more than 350 filed US patents. In Red Hat my focus is on performance and scalability of AI inference infrastructure, focusing on technologies such as vLLM, llm-d, and KV-cache management. My research (PhD in AI) interests include time-series analysis, explainable AI, and scalable machine learning systems.
Optimize long-context AI inference with Red Hat AI 3.5 KV cache offloading, which boosts throughput by 5.8x and cuts TTFT from 60s to 10s on NVIDIA H100 GPUs.
