articles and knowledge

Your model works, but every request is slow and every GPU hour is expensive. Here are five key takeaways behind Inference at Full Throttle, a practical guide from the Scaling AI Systems Series on LLM serving performance. On Kindle and in paperback.

Three focused engineering books from ChatVariety Team for teams running AI in production — fast LLM serving with vLLM, quantization and speculative decoding, reliable GPU cluster operations, and AI unit economics — with a recommended reading order and Kindle and paperback links.

A focused guide from the Scaling AI Systems Series on running GPU clusters for AI: container orchestration, scheduling, observability and failure management. Available on Kindle and in paperback.

Six ChatStick Team guides to building LLM applications in production — RAG, MCP, AI agents, multi-agent systems, observability and security — with what each covers and where to start.