微信安全-内容理解算法工程师-视频号安全方向
Core
Develop multimodal large models to identify prohibited content (pornography, illegal acts, rumors, false advertising) in short videos and build business-specific models for precise violation detection.
Role type
Senior IC multimodal machine learning engineer (content safety)
Builds
Production multimodal large models and business-specific fine-tuned models for short video ecosystem governance
Domain
Social media / Content safety / Multimodal AI
Deliverable
production ML models
Required skills
Multimodal large model algorithms, Instruction fine-tuning (SFT), Preference alignment (RLHF/DPO), LLM/VLM technology stack, RAG, Agent development, PyTorch/TensorFlow, Data processing to deployment pipeline
Preferred skills
Content safety domain experience, LoRA/P-Tuning, Prompt Engineering, Data analysis for model optimization
Technologies
LLM, VLM, RAG, Agent, PyTorch, TensorFlow
Responsibilities
Identify violations in short videos using multimodal models, Train and fine-tune business large models, Develop AI agents to assist human reviewers, Collaborate on data loops from labeling to deployment, Explore frontier technologies for ecosystem governance
Seniority
Senior, hands-on IC
