大模型算法工程师-短视频生态
Core
Design, optimize, and deploy multimodal large language models for TikTok's short video and live streaming ecosystem to understand content, identify gray content, and optimize content distribution.
Role type
Senior Multimodal Large Language Model Engineer
Builds
Multimodal content understanding models, gray content identification systems, and content governance strategies for TikTok
Domain
Social Media / Content Safety / Multimodal AI
Deliverable
production ML models
Required skills
Multimodal deep learning, LLM fine-tuning (SFT/Continual Training), Prompt Engineering, Computer Vision, NLP, Reinforcement Learning, Model deployment & optimization, A/B testing
Preferred skills
Few-shot/Zero-shot learning, RAG, Agent strategies, Content safety domain expertise, Large-scale data processing
Technologies
PyTorch, TensorFlow, Linux
Responsibilities
Design and optimize multimodal algorithms for content understanding and gray content identification; Develop and deploy LLM strategies (CoT, Few-shot, RAG) to reduce issue resolution time; Build RL-based feedback loops for content distribution optimization; Engineer model deployment for high-scale traffic scenarios; Establish and iterate model evaluation systems via A/B testing.