KVCDN Workshop
Best Practices for Pooling LLM KV Cache with FlexKV
Date, time, and room will be added once confirmed.
Talk overview
This workshop explains how FlexKV manages and compresses KV cache across GPU memory, CPU memory, SSDs, and distributed storage. It covers cross-node pooling, asynchronous transfer, namespace isolation, and integrations with vLLM, SGLang, TensorRT-LLM, and Dynamo, drawing on practical enterprise deployment experience.
