[vLLM Office Hours #39] Intro to batch invariant in vLLM – January 8, 2026

Welcome to vLLM office hours! These bi-weekly sessions are your chance to stay current with the vLLM ecosystem, ask questions, and hear directly from contributors and power users.

This week’s special topic: Intro to batch invariant in vLLM

We will start with our regular bi-weekly update from Michael Goin, then welcome Wentao Ye from Red Hat and Bram Wasti from Meta. Wentao and Bram will introduce the core concepts behind batch invariant execution, including what it is, why it is needed, and how it is implemented in practice. They will then dive deep into the reinforcement learning work, with a particular focus on what has been built for TorchTitan, covering the relevant RL theory, key design implications, and results from recent experiments.

Want to join the discussion live on Google Meet? Get a calendar invite by filling out this form: https://red.ht/office-hours