At the 32nd Euromicro International Conference on Parallel, Distributed, and Network-Based Processing (PDP 2024), we presented LARA, a tool that improves latency in stream processing applications, a crucial requirement for real-time video analytics. By using a regression-based resource allocation technique, LARA improves latency by up to 2.8X and delivers over 2X throughput compared to fixed allocation, surpassing the vertical pod autoscaler (VPA) in performance.