We're open source and building in public. Follow along for design decisions, engineering notes, and what we're seeing across GPUs, inference, and production AI systems.