Shaoting Feng
Advisor at Tensormesh
Foster City, California, United States
Summary
🤩
Rockstar🎓
Top SchoolShaoting Feng is an advisor and engineer focused on drastically improving AI inference efficiency, with five years of experience building production-grade systems used by enterprises like NVIDIA and IBM Cloud. At Tensormesh he shipped high-impact optimizations—20× faster KV cache transmission, 2.29× TTFT improvement via dynamic CPU offloading, and 5.49× TTFT gains for multimodal workloads—helping operate >300TB of KV cache and serve over a billion weekly hit tokens. A University of Chicago pre-doc CS student and former EPFL exchange, he blends rigorous research—e.g., practical fairness metrics for network allocation—with hands-on systems engineering. Active on the LMCache team, he’s comfortable taking ideas from prototype to large-scale deployment and balancing GPU/CPU trade-offs in latency-sensitive inference.
5 years of coding experience
Semester Exchange Computer Science, Semester Exchange Computer Science at EPFLPre-Doc Master Computer Science, Pre-Doc Master Computer Science at University of ChicagoBachelor of Engineering - BE Information Engineering, Bachelor of Engineering - BE Information Engineering at Shanghai Jiao Tong University
Github Skills (25)
pytorch10
llama10
inference10
tpu10
amd10
llm10
mlops10
llmops10
xpu10
cuda10
transformer10
gpt10
redistributable7
decimal5
python4
Programming languages (5)
Github contributions (5)
Shaoting-Feng/LoRa-MAC-Layer
Apr 2023 - Feb 2025
MAC Layer for LoRa Communication Protocol
Contributions:1 review, 16 PRs, 18 pushes in 1 year 10 months
Shaoting-Feng/LMCache
Dec 2024 - Mar 2025
Making Long-Context LLM Inference 10x Faster and 10x Cheaper
Contributions:69 PRs, 112 pushes, 25 branches in 3 months
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.