智能座舱AI模型开发_XC

Lead post-training and fine-tuning of core cabin models for speech recognition and multimodal fusion, specifically addressing Code-Switching and dialect-Putonghua mixing to improve instruction parsing accuracy. Design the Agent Function Calling framework to map natural language intent to vehicle control APIs, resolving multi-turn dialogue and context reference challenges. Develop decision models integrating visual, voice, touch, and vehicle state signals to handle cross-modal conflicts and optimize collaborative efficiency. Ensure seamless model adaptation to cabin hardware and software, supporting product R&D, demo validation, and exhibition requirements while leading technical standard formulation. Requires 3-5 years of AI post-training experience with expertise in TensorFlow/PyTorch and multimodal algorithms.