Responsibilities
< /div> Team introduction: Douyin's recommendation technology team is responsible for the full-stack optimization of the recommendation page that accounts for the highest proportion of Douyin's usage time. The underlying model also supports Douyin's other important business scenarios. Our work involves the optimization of large-scale recommendation algorithms, the solution of complex constrained optimization problems, algorithm improvement work in multiple academic fields such as CV/NLP, the design and implementation of recommendation architectures for multiple scenarios, and the complex in-depth study of product data. analysis work. 1. Here, you can come into contact with ByteDance's latest and industry-leading recommendation technology, and experience the full-stack development of ultra-large-scale machine learning systems and recommendation systems 2. You can optimize the graphic and text content in Douyin's recommendation feed by using cutting-edge technology. Core modules include ranking models, multi-objectives, recall, cold start, shuffling, exploration, etc. 3. Current main research directions include: sequential learning, multi-task learning, multi-modal recommendation, sparse target depolarization, and learning to rank, depth graph neural network, etc.
Qualifications
1. Excellent data structure and code skills 2. Have experience in recommendation systems or machine learning, and full-stack development of ultra-large-scale systems Strong interest 3. Excellent problem analysis and problem-solving abilities, full of passion for solving challenging problems 4. Familiar with one or more of machine learning, reinforcement learning, natural language processing, and computer vision, and have knowledge of recommendation systems, computing Applicants with experience in advertising and search engine related fields will be given priority.