Jingren Zhou

Zhou graduated from the University of Science and Technology of China and earned a computer-science doctorate at Columbia University. Before Alibaba, he worked at Microsoft. His SCOPE research addressed the problem of making enormous distributed computations efficient while preserving a high-level interface for developers.[2][3][4]

Alibaba’s current partnership information lists Zhou as chief AI architect and head of the Tongyi Foundation Model Division, alongside his senior vice president position. An earlier Fudan biography describes his leadership of cloud big-data platforms and AI research teams in China and the United States. His responsibilities have moved from the systems that process data at scale toward the architecture and organization of foundation-model development.[1][2]

A database language over a distributed engine

SCOPE combines a SQL-like declarative language with a distributed execution engine. Developers describe the computation without writing its parallel execution plan. An optimizer turns the script into a graph of operations; a job manager schedules those operations across machines and handles failures. Zhou’s invited talk described its use for Bing and other Microsoft services, showing how database optimization and MapReduce-style execution can reinforce each other.[4]

Learning from recurring computations

The SCOPE paper he coauthored examines recurring jobs. Complex user-defined functions and distributed data make accurate statistics difficult to obtain before execution. Reusing information from previous runs can improve the estimates for subsequent instances and produce better plans. The experiments evaluate that approach in a production environment. This work links the performance of individual queries to the efficient use of a shared cluster.[3]