Server Platform (Traffic)

Infrastructure - Observability Team

• Designed, implemented, and safely rolled out Veneur metric pipeline that processes 500M timeseries

• Designed and implemented autoscaling Veneur Aggregator Service that saves $10M+ on annual metric cost.

• Improved the reliability of legacy metric pipeline and fixed the metric loss issue for all Kubernetes pods.

• Developed an in-house error processing and debugging pipeline that are used by 1000+ Airbnb engineers.

• Implemented a customized Kubernetes component that enables distributed tracing for all services.

I was a software engineer intern on Observability team. I helped our team move toward OpenTracing compliant Tracer and stateful stream processing.

• Redesigned the event ingest pipeline to use Flink API that makes the pipeline highly configurable and composable.

• Implemented a customized Flink Executor that executes the event ingestion pipeline.

• Created a generic pipeline that ingests OpenTracing Spans which greatly reduces code duplication.

I was a Software Engineer Intern in ZippyDB team under Data Infrastructure.

• - Created a command line tool using C++ to operate Time To Live data, queue data, and Real Time Counter Service data in a local rockdb key-value store database.

• - Implemented a benchmark tool using Python and C++to measure the performance and limit of ZippyDB distributed database system under the combination of different read/write workloads, different qps on different storage types (disk, flash, tmpfs).

• - Improved the original benchmark workflow using HDFS so that the database preloading time are reduced from 30 minutes to less than 3 minutes .