LLM Algorithm Engineer

We are looking to add a Large Language Model (LLM) Algorithm Engineer in Changsha, China within our EM Labs team. It is a great opportunity to work in a tech-driven company. In a relaxed and friendly environment, our headquarters are in the heart of the city, at Runhe Financial Center, full of interesting and challenging projects. You will be responsible for the production deployment and optimization of LLM environments, including application development and design of distributed deployment solutions based on NVIDIA hardware architecture. Your role will involve leading framework selection and performance tuning for various technologies to achieve high-throughput inference services. You will also build a multi-modal GPU cluster management system to optimize KV Cache storage and loading strategies, enhancing service efficiency for long-context scenarios. Additionally, you will design prompt engineering strategies combined with Retrieval-Augmented Generation technology to improve response accuracy in intelligent customer service and knowledge-based Q&A scenarios. Familiarity with LangChain/AutoGen frameworks is required, and you will lead model fine-tuning using efficient parameter tuning techniques. Developing enterprise-grade internal toolchains and designing external customer systems will also be part of your responsibilities. This position requires a strong technical background and the ability to work independently on complex projects.

Master’s degree or above in Computer Science, Artificial Intelligence, or related fields. 3+ years of professional experience in NLP/LLM projects. Proficiency in PyTorch/TensorFlow frameworks, with deep understanding of Transformer architecture and optimization of attention mechanisms. Familiarity with CUDA programming and NVLink topology design. Experience in NVIDIA chip operator development is a plus. Mastery of development and deployment frameworks such as LangChain, vLLM, and SGLang. Pre-training or fine-tuning of open-source large models. Development of intelligent customer service systems. Construction of knowledge graphs in domains like e-commerce or internet industries.