领悦南京职位，且需要7.20、7.21参加线下面试
1. Key Objectives and Scope of the Position
· As the central Data/AI competence center of BMW China, CIH-6 has the mission to provide user-centric Data/AI interactions with practical innovative attemps, cost-efficiency solutions and graranteed algorithm performance.
· The data scientist in CIH-6 should handle not only traditional data analytics & algorithm approaches, but also cutting-edge technologies (AIGC/LLM e.g.) with code-level prototype build-up capabilities.
2. Major Responsibilities and Degree of Influence
· Discover business values through the entire data pipeline.
· Deliver Data/AI solution based on code-level open-source libraries.
· Deliver Data/AI solution based on major cloud buding blocks.
· Conduct continous research on cutting-edge technologies and explore promising topics in relevant areas.
· Regular progress update with management and stakeholders using data visualization techniques and audience friendly approach.
· Regular knowledge sharing to enable team members.
· Work with team members and stakeholders to validate the technical feasibility of applying cutting-edge technologies in business scenarios.
· Together with DevOps colleagues to accomplish Data/AI function delivery.
3. Qualifications
Education / Degree:
· Ph.D degree in Computer Science, Engineering, Information Technology, Statistics or similar qualification and/or Master in Computer Science, Engineering, Information Technology, Statistics with equivalent work experience.
Knowledge / Skills /Competences:
· Demonstrated experience applying data science methods to real-world data problems.
· Solid knowledge in Data/AI with the capability to determine applicable algorithms/models based on problem statements.
· Code engineering skill to deploy Data/AI models with container technologies.
· Proficient in Python, familiar with python data analytics/AI/ML libs.
· Skills on Data/AI model inference performance optimization with pruning/quantization approaches, and familiar with model optimization tools such as TensorRT.
· Knowledge on heavy-volumn data processing for LLM feeding.
· Proficient in TensorFlow, PyTorch, Megatron, DeepSpeed frameworks, understands various parallel strategies, and has experience in large-scale distributed training.
· Familiar with the fundamental principles and training methods of large models in the industry, such as the three-stage RLHF training for GPT series, LLaMA, ChatGLM, and LoRA fine-tuning, etc.
· Able to design and implement high-performance, highly available backend services tailored for large model requests, ensuring low latency and high throughput data processing capabilities.
· Priority given to those with technical application experience in pre-training, fine-tuning, and reinforcement learning directions for billion-scale large models.
· Good teamwork spirit & conflict resolution skills & creative thinking & problem solving skills.
· Passionate for innovation with a drive to learn new technologies and techniques.
· Excellent verbal and written English.
· Good communication skills in both Chinese and English for coordinating across teams.
Experience:
· Experiences in scripting language Python.
· Expert knowledge in LLM finetuning & optimization.
· Experience in industrilized LLM prompt engineering.
· Experience on AI-driven large-scale distributed system development
· Project, operation experience on DevOps environment and methodology