Worked on the tt-inference-server repository to deliver a hardware-aware optimization feature for Galaxy devices, focusing on model inference performance. Developed a new device model specification for the Galaxy6U, enabling data parallelism with DP=4 and tuning memory cache settings to improve throughput and latency. Integrated the DP4 configuration into the continuous integration pipeline, aligning tests and deployment paths for streamlined release readiness. The work was implemented using Python and machine learning techniques, with an emphasis on model development and deployment optimization. No bug fixes were recorded during this period, reflecting a targeted effort on feature delivery and system integration.