Responsibilities
Team introduction: The ByteDance recommendation architecture team is responsible for the design and development of the recommendation system architecture of ByteDance's related products, ensuring system stability and high availability responsible for optimizing online services and offline data flow performance, solving system bottlenecks, and reducing costs abstracting common system components and services, building recommendation middle platforms, and data middle platforms to support the rapid incubation of new products and empower ToB. 1. Participate in the research and development and optimization of large-scale machine learning online prediction frameworks to support efficient model reasoning for Bytedance-related products such as Douyin 2. Participate in the research and breakthroughs of large-scale feature and sample processing and circulation, model training and scheduling, and other related issues 3. Work closely with recommendation algorithm classmates to analyze and solve the above problems and tasks, and feed back into actual business.
Qualifications
1. Proficient in C/C++ programming in the Linux environment 2. Familiar with operating system knowledge and computer architecture 3. Understand the underlying framework implementation of a certain mainstream machine learning framework (Tensorflow/PyTorch, etc.). Bonus points: 1. Have hands-on experience in deep learning model estimation and training 2. Have implemented custom operators, or optimized existing operator implementations 3. Be familiar with model pruning, quantization, compression and other methods 4. Participated in software and hardware integration optimization work.
