Confirm Re-built deep learning based ID classifier with Facebook's tech stack  Extended ID classification from US only to a global scale by adding a new OCR-based classifier and an ensemble classification layer.  On-boarded several new machine learning engineers

 Utilized AI in computer vision, identity proofing and document authentication by developing innovative deep learning models and advanced data mining workflows. Developed and productionized deep learning models for homogeneous classification, image segmentation and feature extraction.  Won as a team and acquired by Facebook.

 Developed and deployed actionable machine learning models with 200GB datasets to identify student learning patterns, and helped over 35,000 students to improve learning performance.  Spearheaded the use of deep learning models with Python and TensorFlow in educational data mining and outperformed well-known methods by 10% in AUC.  Studied various optimization technologies in an extended Hadoop environment and improved performance by 5 to 10 times on join operations by using colocation and caching strategies.

 Studied the design and implementation details of a Massive Open Online Course system and the workflow of open source software development. Designed and implemented a shared data access framework prototype with Python and Django to encourage more cooperation and communication on the edX platform.

 Designed a classification model with the Random Forest algorithm and C# to identify valid user complaints. This model reduced complaint processing workload by 55% while maintaining extremely high recall on valid complaints. Designed a real-time anomaly detection model using C# to detect Xbox Live operation issues from social network streaming data. This model used IQR threshold methods and demonstrated the ability to detect outage events 25 minutes before official notification system acknowledged.

 Built and maintained a web-based analysis platform with Python to support the work of over 50 traders and analysts.  Optimized computation operations with NumPy and improved time efficiency by 300%.

 Performed I18N & L10N testing on Lotus Notes and related applications to improve production quality. Prompted as Language QA Leader with the responsibilities of coordinating test procedures between testers and developers.