Building Sluice: QoS-Aware Capacity Governance for Self-Hosted LLM Inference

📦 Project: https://github.com/VampiricCyborg/sluice 1. The Problem: When Capacity Becomes the Bottleneck A self-hosted vLLM deployment runs on a GPU pool of fixed size. That's the fact that changes ev...