About the job
Our Recommendation Architecture Team is responsible for building up and optimizing the architecture for our recommendation system to provide the most stable and best experience for our TikTok users.
Responsibilities
Participate in the research and optimization of large-scale machine learning online prediction frameworks, supporting the efficient model recommendation of TikTok's products;
Involve in processing large-scale feature samples, including data flow conversion, model training, and parameter tuning, addressing related issues through research and breakthroughs;
Collaborate closely with algorithm teams to analyze and solve above challenges, and assist with practical business needs.
Qualifications
Minimum
Bachelor's degree or above, majoring in Computer Science or related fields, with experience building scalable system;
Proficient in C/C++ programming in a Linux environment;
Familiar with operating system concepts and computer architecture;
Understand the underlying framework of mainstream machine learning frameworks (TensorFlow, PyTorch etc);
Background in one of the following fields: High Performance Computing, Applied ML models(MLP/Transformer/MoE), ML Hardware Acceleration (e.g., GPU/RDMA) or ML Systems.
Preferred
Experience in deep learning model training and prediction;
Experience in algorithm implementation or optimization of existing algorithms;
Familiar with model pruning, quantization and compression methods;
Experience in software and hardware integration and optimizations.