Skip to main content

Speaker

Xuegang Liang

Xuegang Liang is a core developer of Cambricon’s vLLM MLU project, focusing on model adaptation and inference optimization for the Cambricon MLU platform, particularly Prefill-Decode (PD) disaggregation. He has led optimization efforts for DeepSeek V3, V3.2, V4, and V4.1 on MLU, bringing extensive hands-on experience in model adaptation, inference performance tuning, and system architecture optimization.

Upcoming

Also on the lineup

Is this you?

Claim this profile to keep your talks and details up to date.

A record of public speaking appearances, compiled from event programs. Details can change. Read our trust principles.