Description
The role leads post-training and fine-tuning of seat-cabin core models, including speech recognition and multimodal fusion, and develops an Agent Function Calling framework for natural-language intent mapping to vehicle-control APIs. It also researches visual, speech, touch, and vehicle-state multimodal decision models, resolves cross-modal conflicts, and coordinates model adaptation with seat-cabin hardware and software for product development, demo validation, and exhibition activities. The position requires a bachelor's degree or higher in computer science or AI, 3–5 years of AI model post-training experience, and expertise in TensorFlow or PyTorch, model fine-tuning, quantization, distillation, speech recognition, and multimodal fusion.
