Worked on the RobotecAI/rai repository to optimize the startup performance of perception services by introducing concurrency during initialization. Addressed the challenge of slow readiness for detection and segmentation workloads by implementing multithreading in Python, allowing these services to load in parallel rather than sequentially. This approach reduced startup latency and improved overall resource utilization, enabling faster time-to-value for downstream perception tasks. The solution leveraged ROS2 for robotics middleware integration and included performance instrumentation to validate the concurrency gains. The work demonstrated a practical application of Python scripting and multithreading to enhance system responsiveness in robotics perception pipelines.