I am Kevin Ray Scott (雷凯文 / LEI KAIWEN), a software engineering and artificial intelligence student interested in building reliable intelligent systems that connect models, data, interfaces, and real-world workflows. My recent work focuses on LLM fine-tuning, retrieval-augmented generation, agent harness engineering, reinforcement learning-style optimization, ROS-based robotics interaction, data-driven analytics, computer vision, multimodal AI, and medical AI.

I enjoy turning research ideas into working systems, especially when the system needs evidence grounding, controllable agent behavior, robust failure handling, interpretable evaluation, and practical user interfaces. Across my projects, I try to think beyond isolated models and build complete workflows around data preparation, model adaptation, evaluation, deployment, and user interaction.