Working in Clusters-Core team managing clusters and related resources of the Spark runtime environment.

Key Achievements

• Significantly improved the stability and reliability of the Databricks Cluster APIs by re-architecting the stateful cluster management services to be stateless and replicated.

• Drove the investigation of multiple JVM/MySQL performance issues and implemented optimizations to sustain the fast-growing workload.

• Reduced the operaitonal risks by developing tools, SOPs and best practices for production changes such as schema change, deployment, hotfix, etc.

Lead the cloud team to design, develop and maintain the first commercial graph database as a service platform: TigerGraph Cloud

Software Engineer - GraphStudio Backend, Graph Algorithm, Database Engine, License System and Continuous Integration

I worked as an intern in Blockstore Team of Riverbed Technology.

As project leader, I developed a disk-level prefetch algorithm in C++ to do real-time prediction on which data blocks will be accessed based on current data block access pattern.

The algorithm increased cache hit rate from 5% to 90%, and decreased remote system booting time from 8 min to 4 min.

My work on client pre-fetch was filed for patent application.

I worked as an summer analyst intern in Real-Time Market Data Team of Morgan Stanley.

During the internship, I designed and developed Data Distribution Network Visualization and Configuration Tool as project leader.

This project was developed in Python, jQuery, D3JS and Django, which provides a GUI interface to manage configuration files of clusters with thousands of computers by visualizing the data distribution network, detecting loops and errors, unifying configuration file formats and highlighting critical paths.

The project won most innovative project among 57 intern projects.