Responsibilities
TikTok-Data Video Recommendation Team is responsible for the personalized recommendation algorithms for TikTok's hundreds of millions of global users. Here, you will collaborate with top algorithm engineers in the industry, leveraging your expertise in deep learning, recommendation algorithms, and large models to continuously transform and enhance the TikTok user experience and content ecosystem.
In particular, the local service recommendation team focuses on targeting user experience and transaction scale optimization for lifestyle service content, including hotels, travel, dining, and more. This role aims to pioneer new content and revenue streams for the company.
About TikTok Location Products
Want more jobs like this?
Get jobs in Singapore delivered to your inbox every week.
We creatively connect various products and services related to life through various products such as Points of Interest (POI), videos, LIVE, and search, making users' daily life experiences richer, more unique, and innovative. At the same time, we will also create a business environment which is inclusive and fair, helping businesses, service providers, creators, and other stakeholders to continuously generate more income and improve service efficiency. We firmly believe that through innovation and efforts in life services, we can jointly shape a better and more fulfilling life.
Responsibilities:
This role focuses on recommendation algorithms for international short video-based local services. The work involves optimizing large-scale recommendation algorithms, solving complex constraint optimization problems, improving algorithms across various academic fields such as content understanding, LLM applications, CV/NLP, exploring new business directions, designing and implementing recommendation system architectures for multiple scenarios, and conducting in-depth analysis of product data.
Here, you will have the opportunity to deeply explore the optimization and enhancement of machine learning algorithms and engage with the most cutting-edge recommendation system architectures and large recommendation models in the industry.
The project is driven by technological innovation and aims to revolutionize the longstanding paradigms of recommendation model structures and infrastructure (Infra) by exploring large model solutions in the recommendation domain. These innovations will be applied across various international local service scenarios. The project encompasses research directions such as scaling up recommendation model parameters, cross-modal alignment and unified representation learning (including recommendation, multi-modal content, and natural language), ultra-long sequence modeling, and generative recommendation models, with the goal of systematically upgrading models used in international local services recommendation scenarios.
Qualifications
Minimum Qualifications:
1. Solid foundation in machine learning and programming skills, with in-depth research experience in machine learning, NLP, CV, and proficiency in core algorithms and data structures;
2. Prior experience in generative + search advertising projects (e.g., TIGER generative recall, LRM generative prediction, generative bidding, etc.) is a strong plus;
2. Experience or interest in modeling and aligning multi-modal information such as visual, text, and audio to enhance content understanding and user intent matching, driving the recommendation system toward deeper semantic understanding;
4. Candidates with prior experience in generative recommendation or search advertising projects, or those highly passionate about generative technologies (e.g., TIGER generative recall, LargeRecModel generative prediction, OneRec unified generative prediction, AIGB generative bidding), will be given priority;
5. Publications in top-tier international conferences are a strong plus, including but not limited to KDD, SIGIR, RecSys, ACL, and NeurIPS;
6. Strong analytical and problem-solving abilities, a passion for technology, and enthusiasm for tackling challenging problems and driving innovations.