Da Teng Email and Phone Number

Unified Analytic Platform: Managed cross-time zone teams to deliver customer-facing Analytic products to improve game studio revenue, game player retention, and accuracy of the game campaign.Data Unification Initiative: Managed multiple cross-time zone teams to deliver the centralized solution for Data from ingestion, processing toolsets, unified data store, and self-served backend services to front-end frameworks. The goal is to help any team with an easy plugin to present data insights to end customers.

Tech LeadData Processing Service: deliver an in-house solution for providing self-served capabilities for non-data engineers to write/contribute data pipelines with comprehensive monitoring/alerting solutions. This tool is widely adopted in the company with more than 300 pipelines that are in production. Built and maintained critical time-sensitive customer-facing billing pipelines, designed a dedicated testing process to make sure capture any issues before billing data show to customers. Reduced incidents and frequency and the largest volume pipelines are at 20TB per hour.

I work on the Salesforce IoT Cloud, a massively scalable event processing engine for realtime engagement and analytics. My main focus is that design and implement clusters provision and management for IoT cloud. 1. Provision and manage AWS and Heroku resource, heavily involving Cloudformation, EC2, Auto-scaling and ELB. More AWS and heroku products touched are Route53, SimpleWorkFlow, RDS, S3, Heroku postgres, Heroku kafka, Heroku cassandra and etc.2. Deploy micro-services, manage versioning, upgrading IoT platform properly, scale up and scale down micro-services when necessary. Micro-services that we support use Kafka, Storm, Spark, Cassandra, Kubernetes and so on. 3. Find the best practice to store sensitive and critical information, for example, auth token, client secret. 4. Customize platform when business needs dynamically, involving database migration, kafka migration, Cloudformation stack changes. We are now co-hosting workloads across 10K customers, batch transforming ~100MM rows of user data daily and real-time ingests 7 TB of events daily, ~80 MB/s sustained.

• Developed a face detection and recognition program in Java, implementing machine learning algorithms,e.g., eigenfaces (PCA) and fisherfaces (LDA);• Applied text mining techniques to subtitles to assist the facial identification of actors in TV programs;• Achieved up to 98% accuracy rate on standard face databases, e.g., ORL.