Building Tensormesh: A Conversation with the CEO (Junchen Jiang)

𝗝𝘂𝗻𝗰𝗵𝗲𝗻 𝗝𝗶𝗮𝗻𝗴, co-founder and CEO of ‪@Tensormesh‬ shares why KV cache — the internal memory of large language models, is emerging as the 𝗻𝗲𝘅𝘁 𝗕𝗶𝗴 𝗗𝗮𝘁𝗮 𝗹𝗮𝘆𝗲𝗿 𝗳𝗼𝗿 𝗔𝗜, and how Tensormesh is being built to tackle large-scale inference.

In this conversation, Junchen shares how 𝗧𝗲𝗻𝘀𝗼𝗿𝗺𝗲𝘀𝗵 approaches LLM inference — why performance and cost are often misunderstood, and why memory, specifically the KV cache, sits at the center of the problem.

He also discusses how Tensormesh is being built at the intersection of research, open source, and industry to shape how inference systems are designed, optimized, and scaled.