As a member of one team of the Platform for AI, a.k.a. PAI, within Alibaba Group:1. Helped to build the infrastructure for large scale machine learning by implementing a Parameter Server system to train large models with billions of parameters and integrating into Max Compute – an internal platform for Big Data processing. 2. Worked with PAI TensorFlow, a customized version based on upstream TensorFlow, to optimize the resources allocation, to isolate different CUDA runtimes, as well as to implement an failover mechanism for an Graph Neural Network module.3. Worked on several others projects, including – an online prediction service to design and implement of model deployment/serving/management; an AB Test framework to publish and configure different versions of APIs and distribute request among them in a flexible way.

1. Helps to build Public/Private Cloud Infrastructure with OpenStack technology, focusing on Neutron/Networking.2. Architects and helps to transform Networking Orchestration system for CDN into a fully distributed platform with more intelligence, adaptive capability and higher operability.

1. Leads a Cloud Computing project, use OpenStack as a reference and baseline for Private Cloud Sysstem.2. As a tech-lead, helps with building up a group for indexing building, focused on. anti-spam, junk-detection, page quality; lends some hands to release new word segmentation library and service.

Worked in PDU CC to design and implement software used in networking device OMS1410, an MPLS empowered switching box.

Internship at USB device driver team for Solaris, worked with building HID device driver and power-kit software.