vLLM Inference Meetup Mumbai

A technical vLLM and llm-d meetup on 26 September 2026, exploring scalable LLM inference through architecture deep dives, production lessons and live demos. ## Inside the LLM inference engine Designed for engineers who want to run generative AI systems faster and more efficiently, vLLM Inference Meetup Mumbai goes beyond high-level discussion to examine the engineering behind inference at scale. The programme focuses on the architecture and optimisation techniques that help teams make better use of GPU infrastructure, increase token throughput and move LLM services into production. Attendees will hear directly from maintainers and core contributors working across the vLLM and llm-d ecosystems, alongside industry practitioners with experience deploying production AI systems. Sessions will combine project updates and technical talks with practical demonstrations of real-world workflows. ## Agenda highlights The planned programme includes: - An introduction to vLLM and an update on the project’s latest development - Speculative decoding and its role in accelerating AI inference - A journey from evaluation to production, including how to build a managed vLLM service with llm-d - Lessons from teams deploying and serving LLMs in production - A Sardeenz hands-on lab covering GPU virtualisation and multi-model deployment on a single GPU - Live demos, opening remarks and opportunities to exchange ideas with speakers and fellow engineers The agenda is described as subject to change, leaving room for further technical content and community contributions. ## Who should attend? This meetup is particularly relevant to vLLM and llm-d users or contributors, machine learning and infrastructure engineers responsible for inference and model serving, and platform teams operating GenAI workloads in production. It will also suit developers investigating efficient inference across local environments, cloud infrastructure and Kubernetes.… Register for this event