Worked as a core software engineer in a geospatial big data company, which constructs data pipelines to process satellite images, and leverages artificial intelligence to get some insights on a global scale.

Designed and built a data monitoring platform, where data monitors and alerts can be visualized, configured, and set up along the geospatial data pipeline to improve data reliability and quality.

Upgraded database schema on PostgreSQL for time-series data storage, set up alert rules based on historical data analytics using Grafana, and realized the monitoring of image count, delivery delay, and ingestion delay.

Dockerized the data monitors, streamlined the deployment and management of Kubernetes applications using helm, and automated the whole process with continuous integration(CI/CD) on Jenkins.

Deployed the monitoring platform into production, and reduced the time delay of anomaly discovery by 70%.

Collaborated with a team of 4 and designed an in-pipe low-power usage leak detecting swimming robot based on Arduino 101 and sensors(gyroscope and microphone).

Implemented several classification models(e.g. SVM, Bayesian, NN) and achieved leak detection accuracy of 90%, which helped Enbridge(North America leading energy delivery company) save millions of dollars in labor cost.

Published paper on American Control Conference (Jun 27-29,2018, Milwaukee, WI) as first author.