混元大模型数据策略产品（代码方向）（北京）
Core
Define code and agent data categories for LLM training (SFT, RL), establish data quality standards, and build data acquisition/annotation/synthesis pipelines to improve model performance.
Role type
Senior IC data strategy product manager (code/LLM)
Builds
Production data pipelines, evaluation frameworks, and training datasets for large language models
Domain
Artificial Intelligence / Large Language Models / Code Intelligence
Deliverable
production ML models
Required skills
LLM training strategy, data quality assessment, data synthesis, experimental design (ablation studies), programming proficiency, data analysis
Preferred skills
Experience with code agents, familiarity with developer toolchains, ability to read training logs
Technologies
SFT, RL, data annotation systems, data synthesis pipelines, developer toolchains
Responsibilities
Collaborate with algorithm teams to identify data gaps across training stages; Establish quantifiable data quality metrics covering diversity, correctness, and difficulty; Build and iterate data acquisition, annotation, and synthesis workflows; Validate data impact on model performance through experiments.
Seniority
Senior, hands-on IC
