Created Near Real Time Story Pin Selection system to filter low quality story pins and select the best pins for users in different countries by using Flink, Kafka, S3 and Rockstore

Developed Pin Selection batch pipeline to deliver pin features, validate pin quality, ranking and localization to Pinterest serving platforms- Homefeed, Related Pin and Search by using Hadoop.

Created user engagement reporting pipeline to provide weekly and yearly engagement trends and insights on user interest categories for Pinterest’s business partners by using Python and Hive

Developed indexing pipeline and validation framework for products to serve merchants’ product feeds into shopping ads index by Java

Developed a real time risk analysis system to process 300,000 records market data per second by building up an AMPS high-performance client and generating triggers based on specific events using Python and Vertica

Improved the testing and deployment process by creating unit tests and integration tests for each features and a pipeline for auto-testing and deployment using Bamboo and Ansible

Developed an alert system to automatically detect anomaly on data pipeline to exchanges based on statistical analysis using historical data by Python, Pandas, NumPy and Bash

Increased the performance by 20% to process 100,000 rows of records per second real-time stream by C++, multi-thread and SQLite

Reduced human errors and time to recover order execution system by 25% with automation alerting system

Received a recommendation letter from Innolux Corporation after presenting the optical principles of LCD panels to other product manager and engineer interns and mentors