大模型训练（今日头条）-CQC
Core
Design, train, and evaluate large language model (LLM) safety strategy models for Toutiao's content moderation and platform governance.
Role type
Senior LLM Training Engineer (Safety & Governance)
Builds
Production safety strategy models, RAG applications, and prompt engineering workflows for content moderation.
Domain
Internet / Content Safety / Large Language Models
Deliverable
production ML models
Required skills
Large model training (RM, RL), NLP, Computer Vision, Prompt engineering, RAG application design, Model fine-tuning, Inference optimization
Preferred skills
Experience with CLIP or Diffusion Models, Programming competitions, Multi-modal generation models
Technologies
Python, Java, Go, JavaScript, C, C++, SQL, NoSQL
Responsibilities
Design model architecture and data pipelines for safety strategies; Collaborate with business teams to optimize model capabilities based on metrics; Develop and maintain prompt workflows for content governance; Explore and implement RAG solutions for specific business scenarios; Monitor and iterate model service quality standards.