Responsibilities
Team introduction: TikTok content ecological algorithm team is based on the platform's massive users and short video content, using multi-modality, LLM/MLLM, NLP&C V and other technologies are responsible for content layer operations such as analysis, processing, and generation of various types of content. Content forms include but are not limited to short videos, graphics, hotspot pages across the entire network, customer service conversations, etc., and produce content understanding features, large model generation intermediate pages, hotspot discovery & understanding capabilities across the entire network, and intelligent customer service systems. At the same time, the supply side of TikTok provides personalized recommendation capabilities for creative inspiration across all platforms. Here, there are hundreds of languages, and technologies such as NLP/LLM face multilingual challenges. Here, there is a massive amount of short video content, and MLLM and multi-modal technology have more application scenarios. The platform has a large number of users, and various businesses such as local life and search are in a period of rapid development, with a large number of actual implementations and application scenarios. 1. Participate in the research and development work related to NLP/CV basic algorithms in the TikTok business, deeply understand the business, solve and follow up on front-line business problems 2. Work closely with the product operation team to continuously iteratively optimize algorithm effects in business directions such as hot spot mining and local life, and achieve business goals 3. Combined with business scenario challenges, track research results in cutting-edge fields, and promote the in-depth application of technological innovation in business scenarios, including but not limited to hot spot discovery/information extraction and structuring/multimodality/large models/text generation/retrieval and correlation and other algorithm fields.
Qualifications
1. Have solid machine learning foundation and mathematical skills, NLP/CV/multi-modal/large model and other related backgrounds are preferred, have the ability to independently tackle key problems and solve practical problems, and experience in practical application of large-scale scenarios is a plus 2. Be familiar with mainstream text/visual pre-training models, and have at least engaged in algorithm model development and implementation business in any direction (hotspot discovery/information extraction and structuring/multimodality/large model/text generation/retrieval and correlation) be proficient in using mainstream machine learning frameworks and model training frameworks 3. Have strong technical curiosity, self-drive and enterprising spirit, and be able to pay attention to and learn industry best practices in a timely manner have strong modeling capabilities from business problems to algorithmic models, have a strong sense of cooperation, be good at cross-team communication and promote the solution of practical problems.
