Serve Your Own LLM: vLLM & SGLang, End-to-End

This course, "Serve Your Own LLM: vLLM & SGLang, End-to-End," offers 12 live lectures starting in November 2026. It covers tracing requests from chat to GPU, building latency and throughput benchmarks, and tuning memory, batching, prefix caching, quantization, multi-GPU, and mixture-of-experts serving. The curriculum also includes production aspects like routers, Kubernetes, metrics, and cost per token, culminating in a capstone project to compare vLLM and SGLang.