Hui Zhang
Serving LLM/VLM @ByteDance
- Role
- Senior Researcher Architect, Tech Lead at ByteDance
- Location
- San Jose, CA, US
- LinkedIn followers
- 500 followers
About Hui Zhang
10+ years of experience in the infrastructure innovation for AI, HPC, Bigdata w/ HW/SW co-design.10+ Papers, 30+ Patents in the related areas.Currently serving LLM/VLM in Bytedance Seed Infrastructure team
Experience
Senior Researcher Architect, Tech Lead
Jan 2022 — Present · San Jose, CA, US
MLSys- Building ByteDance core Inference EngineInfra for AI- GPU Sharing: CUDA interception to improve GPU utilization from 34% to 63%- NCCL Optimization: Memory pooling to improve CC operators\' throughput- GPU Filesystem: GPU-initiated direct SSD access for faster multi-tier KV cache managementBigData- Led a HW/SW co-design project to accelerate Cloud-Native OLAP DBMS with HW offloading; industry-first deployment, decreasing TCO > 20%Storage- Key-Value storage optimization (similar to RocksDB)- CXL memory technology exploration (EuroSys\'24 best paper final runners(1 out of 2))- TL of Instance Storage project (similar to AWS Nitro Card for Instance Storage)- TL of DPU Stability Enhancement project (innovative versioning and auto-testing)
Education
Beihang University
Bachelor's degree, Technology and Apparatus of Measurement and Control
2007 — 2011
University of Maryland
Doctor of Philosophy (Ph.D.), Computer Engineering
2011 — 2018
Find verified contacts for anyone on LinkedIn
Unifers gives sales teams verified emails and direct dials, enriched profiles, and outreach that lands in the inbox.
Free plan included · No credit card required
This profile is compiled from publicly available professional sources. Unifers is not affiliated with or endorsed by LinkedIn. Request removal of this profile.