1. Be responsible for building an efficient, reliable and cloud-side collaborative in-depth learning cloud native AI platform.
2. Participate in the research and development of important issues such as intelligent annotation, model training, model inference engine, feature engineering, resource scheduling, and improve the usability of the platform;
3. Deeply optimize gpu virtualization, storage optimization, rdma and other core technologies in the container scenario to improve the efficiency of the platform
4. Focus on the forefront of the industry and continue to optimize the platform.

Job requirements
1. Proficient in one of python, java, golang and c++programming languages, and have practical project application experience;
2. Understand the mainstream in-depth learning framework: tensorflow, pythory, paddlepaddle, horovod, etc., and understand its core working mechanism;
3. Familiar with mainstream inference engines and acceleration tools, such as Triton, TensorRT, etc
4. Be familiar with Kubernetes, have a certain scale of production practice experience, can independently analyze and solve the problems of various components of Kubernetes, and have secondary development experience is preferred;
5. Be able to analyze and locate the performance bottlenecks of training and serving tasks, and formulate corresponding optimization measures;
6. Experience in heterogeneous computing platforms is preferred, and practical experience in gpu programming, gpu virtualization, computing and storage separation architecture optimization is preferred.
7. Those who have experience in AI platform research and development based on kubernetes are preferred, and those who are familiar with MPI/NCCL are preferred;