AI Slop
Coding assistants feel effortless. The evidence says they make us write more code and work slower, and the dependency that builds only benefits one party.
Writing
Writing about LLM inference, long-context serving, GPU and CPU architecture, and the systems work that sits between a model and the hardware it runs on.
2026
KV Store: paying memory to avoid recompute
Without a KV cache, every new token re-attends over the whole prefix. With one, you trade that compute for GPU memory — about 40 GB of it on Llama 3 70B at 128K context, and PagedAttention decides how much of that you waste.
Starting a blog
Why I am writing things down in public, and what I plan to put here.