Job duties：
Participate in RLHF/PPO training and SFT of LLMs to optimize performance in motor control chip technical Q&A and customer support.
Design reward models and preference alignment strategies to improve response professionalism, accuracy, and safety.
Build a domain knowledge base for motor control chips (datasheets, application notes, FOC docs, FAQs) through cleaning, structuring, and vectorization.
Design and optimize RAG pipelines to improve retrieval recall and response quality.
Establish automated evaluation systems, conduct performance analysis and bad case reviews to drive iteration.
Track emerging technologies (LLM, RAG, RLHF) and explore applications in chip industry knowledge services.
Job requirements：
Bachelor's degree or above in AI, CS, Data Science, Math, Electronics, Software Engineering, Automation, or related fields.
Proficient in Python; familiar with PyTorch/TensorFlow.
Understanding of Transformer architecture and mainstream LLMs (Qwen, ChatGLM, LLaMA).
Basic knowledge and strong interest in RLHF, RAG, and Prompt Engineering.
Good paper-reading skills and self-motivation.
Apply
