Tianming Lu
R&D Engineer, eBay AI Platform Division
Tianming Lu, graduated from Nanjing University, has many years of R&D experience in big data and AI. He is now working in eBay AI platform department, and has participated in the research and development of unified feature service, model orchestration optimisation, model deployment process optimisation, model service performance optimisation, etc. He is mainly responsible for improving the usage of computing resources of model service. Currently, he is mainly responsible for improving the usage rate of model service computing resources. He is committed to improving the efficiency and performance of AI model service, and providing enterprise-level high-performance model inference platform.
Topic
Building an Enterprise-Class High-Performance Model Reasoning Platform
With the rapid development of AI technology, enterprises are increasingly relying on machine learning models. However, effectively applying these models to real-world business scenarios requires the support of a powerful inference platform. As eBay's machine learning platform department, we need to ensure the efficient management of models and fast on-line iteration, and at the same time, continuously improve the performance of online inference services to meet the needs of business scenarios. This talk focuses on some of eBay's engineering practices in model inference service optimisation, and introduces our work in model service routing, inference performance optimisation, and compute resource optimisation.
