Thursday, November 12, 2026
3:50 PM – 5:20 PM PST
SGLang Summit Workshop by Crusoe: Serve, Measure, Tune: SGLang on Crusoe Managed Kubernetes
About this event
A 90-minute hands-on workshop at SGLang Summit: deploy an SGLang endpoint on Crusoe Managed Kubernetes with KServe, benchmark an open-weight model under concurrent load and tune KV cache precision and scheduling. Approval required, sold out with a waitlist.
Presented by Crusoe engineers with LMSYS Org at Fort Mason, for intermediate to advanced developers and infrastructure engineers. Attendees deploy an SGLang inference endpoint on Crusoe Managed Kubernetes using KServe, benchmark an open-weight model under heavy concurrent load, and tune settings such as KV cache precision and request scheduling to see the trade-offs. Crusoe supplies the GPU cluster.
Bring a macOS or Linux laptop with uv, kubectl and git (Windows users can use WSL2 or pair up). The room seats 45 first come, first served. Free but approval is required and an SGLang Summit pass (Full or Expo) is needed. Sold out with a waitlist when checked. 3:50 to 5:20 PM Pacific.
Organized by
Community events for SGLang, an open-source high-performance inference framework for large language models maintained by LMSYS Org.