1. Led/participated in large model application projects, built high-performance RAG system, optimized the accuracy and efficiency of retrieval and generation modules;
2. Design model tuning schemes (LoRA/P-Tuning/ full parameter tuning, etc.) to improve the performance of the model in the vertical field;
3. Explore large model compression and inference acceleration technologies to promote model industrial deployment;
4. Build innovative solutions such as domain knowledge enhancement and multi-modal data fusion;
5. Tracked the industry technology dynamics (such as Agent/ long context optimization), and promoted the team's technology upgrade.
1. More than 2 NLP/ large model project experience, leading RAG system development (familiar with LangChain/LlamaIndex and other frameworks)
2. Proficient in large model tuning technology chain: data cleaning - prompt engineering - efficient parameter fine-tuning - evaluation alignment
3. Master PyTorch/TensorFlow framework and have distributed training/multi-card parallel practical experience
4. With 100 billion parameter model pre-training/industry large model landing experience is preferred
5. Have good communication skills and self-drive, can effectively lead the team to carry out technical research and development work, and ensure the timely and high-quality delivery of projects.
Hiring manager and Recruiter should collaborate to create the relevant verbiage.