微信安全-内容理解算法工程师-电商入驻审核
Core
Develop multimodal content understanding and violation detection algorithms for e-commerce merchant onboarding, utilizing LLMs and VLMs to automate risk identification for prohibited goods, fake brands, and compliance issues.
Role type
Senior IC machine-learning engineer (multimodal content safety)
Builds
Automated e-commerce merchant onboarding review systems and business-specific large models
Domain
E-commerce / Content Safety / AI
Deliverable
production ML models
Required skills
Multimodal content understanding, Large Language Models (LLM), Vision-Language Models (VLM), Instruction Tuning (SFT), Preference Alignment (RLHF/DPO), RAG, Agent development, PyTorch, TensorFlow, Model deployment, Data analysis
Preferred skills
E-commerce merchant review experience, LoRA/P-Tuning, Prompt Engineering, Knowledge Graphs
Responsibilities
Build multimodal algorithms for identifying prohibited goods and compliance risks; Train and fine-tune business-specific large models for vertical violation detection; Design human-AI collaboration mechanisms to boost review efficiency; Optimize the engineering architecture of the review system; Establish model evaluation and monitoring loops; Explore frontier technologies like RAG and Agents for review scenarios.
