Samuel Shen

I'm currently located in the Bay Area, working on KV Cache optimizations at Tensormesh.

I enjoy working on systems for ML.

I graduated from UChicago in 2025 and really enjoyed my time there. I've had the rare opportunity of learning how to develop good software through maintaining the open source project LMCache.

updates

May 2026 Speaker at NVIDIA Dynamo After Hours in SF about LMCache MP Mode architecture.
May 2026 Speaker at Unlocked Conference in Seattle about optimizing the storage layer for KV Cache.
April 2026 Release LMCache Agentic Traces on HF.
March 2026 LMCache featured in NVIDIA GTC Keynote!
March 2026 Redis from 0.3 GB/s to >10 GB/s for KV Cache workloads w/ a new Native Connector system — blog.
February 2026 Pretty happy with this refactor: GPUKVFormat :) — PR.
January 2026 P2P distributed storage in LMCache — blog.
December 2025 EVICPRESS on arXiv!
December 2025 LMCache technical report on arXiv!
November 2025 Host LMCache Booth at KubeCon in Atlanta.
October 2025 CoreWeave and Cohere collab to repurpose CAIOS for inference acceleration — blog.
July 2025 LMCache hits 2k stars and joins PyTorch Foundation.
June 2025 Graduate :)
May 2025 Build LMBench.
April 2025 Join Tensormesh as first engineer!

Samuel Shen

links

blog →

updates