Worked on the neuralmagic/vllm repository to introduce a deprecation warning for the lora_extra_vocab_size parameter in the LoRA configuration, preparing users for its removal in future versions. This update aimed to improve API clarity and maintainability by signaling the phase-out of extended vocabulary support, allowing users to adapt their implementations ahead of breaking changes. The work involved Python backend development, with careful attention to clear deprecation messaging and alignment with the product roadmap. No major bugs were addressed during this period, but the contribution demonstrated a focus on forward-compatibility, code quality, and effective communication through documentation and release notes.