Andrew Ng releases new course on fast LLM inference with Cerebras Wafer-Scale Engine, enabling real-time agentic workflows and latency-sensitive applications.