Shuqiang Zhang

2023 to present: Pytorch Distributed/HPC comms libs/High Speed and low latency networks 2021 to 2023: Network.AI team: intersection between computation and networking, building scalable and high-performant systems for Facebook's growing AI workload. 2016-2021: WAN controller team: Building Facebook's global backbone Networks controlled by software.

Developing an application-centric enterprise WAN solution by chaining applications with various network services/functions end-to-end, based on real-time data analytics. Key components involved: OpenStack, OVS, ONOS/ODL, vSphere, SD-WAN, and WAN optimization, etc.

R&D on system design and implementation to achieve optimized trade-offs among Bandwidth, Storage, Computation, and Latency in the context of WAN optimization. Asynchronous Event Programming, Multi-Threading.

Proposed and implemented optimization models for service-centric traffic engineering and cache orchestration in SDN-enabled information (content)-centric networks and CDN. The proposed algorithm reduced maximum link load by 68% for network service providers as well as being scalable to large number of services. Languages/Platform: Python, NetworkX, CPLEX, discrete-event driven simulations.

Proposed and implemented algorithms to design a converged metro network. Reduced fiber deployment cost by up to 70%, and achieved high wavelength switch utilization. Evaluated control-planes solutions (SDN/OpenFlow, GMPLS, etc.) for the metro network. Developed a software with GUI using PyQt. Languages/Platform: Python, PyQt, NetworkX, CPLEX, etc.

Research on multi-layer transport network architecture, software-defined networks, network virtualization, hybrid packet/circuit switching, inter-data center networks. 20+ publications in top journals and conferences, e.g., JSAC (Journal on Selected Areas in Communications), JLT (Journal of Lightwave Technology), JOCN (Journal of Optical Communications and Networking) etc. Best paper recipient at ANTS 2011. Languages/Platform: C/C++, Python, Java, Matlab, OPNET etc.

Research on energy-efficient time-aware traffic grooming, multicast protection under scheduled traffic, and application of game theory on multi-domain optical networks.