大模型推荐算法工程师（交互式推荐方向） - TikTok算法
Core
Designing and training intelligent agents for intent understanding in recommendation systems, building programmable control capabilities for recommendation links, and exploring end-to-end generative recommendation (LLM4Rec) for TikTok's short video product.
Role type
Senior IC large model recommendation algorithm engineer (interactive recommendation)
Builds
TikTok short video recommendation and search algorithms
Domain
Social media, large language models, generative AI
Deliverable
production ML models
Required skills
Deep learning frameworks, large language model training and inference (pre-training, fine-tuning, alignment, decoding), intent modeling, experimental design, data structure and algorithms
Preferred skills
LLM4Rec, Agentic Recommendation, semantic ID representation, multi-agent systems, reinforcement learning
Technologies
LLM, Multi-Agent, Agentic RL, Semantic ID, DPO, PPO, GRPO
Responsibilities
Define intent spaces and construct instruction data for agent modeling; build and validate programmable control capabilities for recommendation links including parameter tuning and active interest distribution; explore end-to-end generative recommendation with personalized generation and decoding strategies; establish evaluation systems and automated benchmarks for intent recognition and user satisfaction; track and implement cutting-edge research in generative recommendation and agentic recommendation.
Seniority
Senior, hands-on IC