Senior TL at Ads Pacing and Auction team to drive ads auction area which involves dozens of engineers and data scientists. We are building and improving the auction platform to help ads supply teams to monetize Meta’s family of apps. Specifically, our focus includes speeding up new ads supply releases, maximizing the tradeoff between revenue and engagement, and maintaining auction integrity.

Tech lead in PGP (Product Growth Platform) ranking infra area. Our mission is to build stable, reliable and sustainable ranking infrastructure to use machine learning technology to maximize the value for FB2U (Facebook-to-user) communications cross multiple channels (like feed, notification and so on).

* FB2U has been extend from Facebook blue app to family of apps (e.g. Instagram, Workplace, AR/VR and so on).

• Worked primarily on feed ads autoML framework to offer model training automation. Built the tool to automate model training process for all feed ads ranking engineers.

• Implemented a data blackout tool for both production and QRT workflows and realized automatic workflow killing with polluted data.

• Developed a new algorithm with higher robustness and stability for red-green target recognition and achieved 10% higher recognition accuracy compared with existing algorithms (used C++, OpenCV and MATLAB)

• Improved the algorithm for reflective circle target recognition in mobile webcam calibration