混元大模型数据策略产品（Stem/Reasoning方向）（北京）
Core
Define and iterate data strategies for STEM and Reasoning capabilities in large language models, translating model evaluation insights into executable data requirements.
Role type
Senior IC data strategy product manager (LLM STEM/Reasoning)
Builds
High-quality training datasets (real and synthetic) for base model development
Domain
Artificial Intelligence / Large Language Models
Deliverable
production ML models
Required skills
Data strategy formulation, STEM/Reasoning domain expertise, data pipeline design, data quality assessment, script writing for data processing, statistical analysis, model evaluation analysis
Preferred skills
Experience with base model training data, engineering capabilities, understanding of data synthesis and curriculum learning
Technologies
LLMs, Agents, Data synthesis frameworks
Responsibilities
Align with algorithm teams to define data goals based on model evaluation and badcase analysis; Design end-to-end data production workflows including collection, cleaning, annotation, and quality control; Establish quantifiable data quality metrics and track training gains; Innovate data production methods using LLMs and automated workflows.
