A practical two day technical course for participants who are new to vLLM and want to deploy large language models efficiently in production-oriented environments. The agenda progresses from model serving and runtime architecture through inference optimization, continuous batching, configuration, distributed serving, and scaling. The course focuses on the practical decisions required to achieve high throughput, predictable latency, and reliable operation while maintaining compatibility with modern AI application stacks.
You will learn how to serve open-weight language models with vLLM, configure the inference engine, and expose models through OpenAI-compatible APIs. You will understand how PagedAttention, continuous batching, prefix caching, quantization, and parallel execution improve inference efficiency. You will also learn how to configure, benchmark, scale, monitor, and operate vLLM services for demanding production workloads.
*We customize the course outline and content to your specific needs and relevant use cases.
Module 1: vLLM architecture and model serving fundamentals
Module 2: OpenAI-compatible model serving
Module 3: PagedAttention, batching, and memory efficiency
Module 4: Inference configuration and optimization
Module 5: Multi-GPU serving and parallelism
Module 6: Scaling concurrent inference workloads
Module 7: Benchmarking, monitoring, and performance tuning
Module 8: Production deployment and operational readiness
Hands-on learning with expert instructors at your location for organizations.
Master new skills guided by experienced instructors from anywhere.
Can't find a suitable date? Get in touch and we'll arrange one that works for you.