Ping Yu Email & Phone Number

Mountain View, Ca, Us

- TLM for Google Edge AI runtime, responsible for building an unified ODML runtime stack that enabling GenAI for on device experiences, across mobile/web/embedded devices, formally known as (TFLite/TFJS/TFLM).- TLM and founding member TensorFlow.js: Google WebML solution. Grow the project at 3x rate since launch in 2018, now with 500M+ CDN download, 18K github stars.- In depth experiences with model conversion, optimization and execution pipeline across CPU/GPU/NPU.

Mountain View, Ca, Us

TL for Google Attribution FEGoogle Attribution is a new initiative that focuses on consolidating all attribution features across allGoogle Ads products, by adding new cross-channel, cross-device, multi-touch attribution for alladvertisers and making the conversion credit available for reporting and bidding. Its target audiencespends over $40B on Google Ads.- Led the team of 6 engineers, successfully launched Google Attribution 360 and GoogleAttribution Free, which target the large enterprise and small business respectively.- Led the continued development of both products and drove the adoption ofdata-driven attribution by 10 times for AdWords Users.

Leads Intridea Product designs: SocialSpring Social Network and Stream PlatformActively engaged in Open Source Community with releasing and supporting several popular Ruby Gems.Designed and implemented automatic deployment and monitoring system for all of our clients. Key Client Projects- RentPath: Led a team of 3 members, building a cloud-based asset service(image/document/video). This service provides millions of contents for several leading apartment searching sites.- Agilysys: Led a team of 10 members, building a revolutionary SAAS POS system. Combining the latest web technologies with QT platform, this system provides the customer with real-time inventory update, offline transaction processing, and comprehensive sales reporting.- GameStreamer: Led a team of 12 members, built a highly complicated B2B digital gamedistribution and social platform using Ruby on Rails.

Bethesda, Maryland, Us

- Led client application development. Designed, and implemented RView - a Swing-based javaapplication that provides real-time charting, various reports. - Enhanced Riorey DDOS attack detection and filtering algorithm for TCP and UDP traffic usingprotocol micro-behavior analysis in C and C++.- Improved Linux kernel device driver for Intel E1000 Ethernet chipset. Increased packetthroughput by three times.- Owned both server and client end of the iCARE System. This system monitors all deployedRiorey DDOS prevention devices, provides reports for attack patterns, plots attack geographiclocations using google map api, and utilizes Ajax for highly interactive web UI.

Us

- Technical Lead on the OCS CEI group. Responsibilities include requirement analysis,framework design and coding, as well as directing the activities of other engineers. Handledleadership aspects of the project including scheduling, team management, and negotiationwith management and system groups. OCS CEI is a standalone Java application utilizes XML,ILOG, and Swing.- Lead developer for the server performance monitoring and automated stress-test system aswell as optimizing server performance by eliminating inter-thread contentions, reducingmemory leaks, and solving database deadlocks with JVM profiling tools.- The Owner of OnDemand Path Management module in Corvis Unified TMS platform suites.The module is implemented as EJB and Java Beans on WebLogic Application Server.- Designed and implemented channelized switching and automatic wavelength contentiondetection algorithm for Corvis legacy systems using Java and C++.

M.S., Computer Science

Bachelor Of Science (B.S.), Computer Science