Azure Hardware Health Telemetry Ingestion Service

• Built large scale distributed telemetry ingestion service to monitor Azure fleet health.

• Built ingestion/data ETL pipelines with Azure Event Hubs, Service Fabric, Synapse (Spark), Data Lake and Kusto.

• Built scalable/flexible infrastructure as code for multi tenants/regions.

Azure Hardware Health Alerting Service

Built service sending Out For Repair (OFR) requests for sick nodes to datacenter ticketing system.

Built alerting platforms, supporting end-to-end node OFR workflows, with customized real-time/analytical alerts.

AI Security Research on Computer Vision and Object Detector

• Proposed an optimization method to craft adversarial examples to attack YOLO object detector with a success rate of 100%.

• Put forward a robust physical adversarial sticker generation method based on an extended Expectation over Transformation with a success rate of 0.72. Recognized and verified the effectiveness of a brand new form of “Off-Target Watermark attack”.

1. Designed an arrangement of RGB-D cameras in parking garage, an auto-parking scenario.

2. Implemented a camera fusion algorithm to handle RGB-D depth data and generate obstacles expression in general coordinate.