Responsibilities
1. Responsible for the research and development of the internal machine learning platform training framework (including data preprocessing/training/inference), serving scenarios such as advertising, recommendation, and search 2. Responsible for the design and development of real-time high-performance estimation systems, such as operator fusion, compilation optimization, model quantization, mixed precision, heterogeneous hardware acceleration, etc. 3. Responsible for performance optimization and architecture upgrades, and continuously improve data preprocessing/training/estimation performance 4. Work closely with algorithm engineers to jointly optimize algorithms and systems for key projects.
Qualifications
1. Proficient in C++, Python and other programming languages those with experience in GPU programming, compilers, distributed computing, high-performance networks, etc. are preferred 2. Familiar with at least one deep learning framework (Tensorflow/Pytorch/MXNet or other self-developed frameworks), and have in-depth research on its underlying principles 3. Familiar with common inference optimization techniques, such as operator fusion, model quantization, mixed precision, etc., and those with relevant work experience are preferred 4. Familiar with common machine learning and recommendation algorithms, such as CNN/RNN/LR/SVM/RF/GBDT/FM/DeepFM/DCN/xDeepFM, etc. 5. Have the ability to solve problems independently, and have good teamwork awareness and communication skills 6. Have faith in technology and pursue perfection. Bonus points: 1. Strong curiosity, love technology and have in-depth research in specific fields 2. Have experience in product development directly facing users 3. Have experience in large-scale distributed system development.
