Responsibilities
Team introduction: Responsible for the search algorithm innovation and architecture research and development of Douyin, Toutiao and other products. We use cutting-edge machine learning technology for end-to-end modeling and continue to innovate and make breakthroughs. At the same time, we focus on the construction and performance optimization of distributed systems and machine learning systems, from optimization of memory and disk to the exploration of algorithms such as index compression, recall, and sorting. We also apply RAG technology to realize the potential value of AI to hundreds of millions of Douyin users. Provide students with ample opportunities to grow themselves. The team's main directions: 1. Exploring cutting-edge retrieval technology: across many genres such as video, live broadcast, graphics, group buying, etc., from basic NLP technology to recall based on multi-modal understanding, user behavior understanding, Embedding similarity, etc. 2. Exploring large-scale sorting technology: Based on the original Bert and large-scale sparse models, explore and implement the ultra-large-scale autoregressive model Scaling Law 3. Explore the ultra-large-scale AI search RAG engine: fully tap the potential value of Douyin's massive traffic, and build an ultra-large-scale, multi-agent collaborative AI search overall architecture to meet potential user value 4. Large-scale streaming machine learning technology: ultra-high throughput real-time data flow, streaming large-scale machine learning, so that more personalized searches can understand you better 5. Architecture for hundreds of billions of data scale: There is in-depth research and innovation in all aspects from large-scale offline computing, distributed system performance and scheduling optimization to building high-availability, high-throughput and low-latency online services. 1. Build Douyin AI search and build an LLM search engine with complete RAG functions 2. Optimize all aspects of LLM search, such as online retrieval and large model reasoning 3. Design and build offline personalized AI search capabilities to achieve more accurate questions and answers for users 4. Continue to build a more flexible, reasonable, and powerful architecture to ensure that the rapidly developing business can be efficiently achieved.
Qualifications
1. Have a solid foundation in computer theory, excellent coding skills, excellent data structure and basic algorithm skills 2. Have a strong sense of responsibility, good communication skills and cooperative spirit, and be proactive 3. Have at least one of the following experiences: recommendation engine, search engine, advertising engine, RAG engine, RAG open source framework. Bonus points: 1. Applicants who are familiar with performance tuning in the Linux environment and have experience in large-scale C++ online service development are preferred 2. Familiar with the theory of distributed systems and have experience in distributed databases , Applicants with experience in research and development of distributed computing frameworks are preferred 3. Applicants who are familiar with the basic principles of machine learning and have actual experience in GPU inference optimization are preferred.
