Zhuoran Zhao (赵卓然)

I am a Research Scientist at Facebook, Menlo Park. Prior to joining Facebook, I received my Ph.D. degree in Electrical and Computer Engineering (ECE) from the University of Texas at Austin in 2019, where I worked in System-Level Architecture and Modeling (SLAM) research group under the supervision of Professor Andreas Gerstlauer. I received a B.S. in Electrical Engineering from Zhejiang University in 2012.

My current research interests mainly include Machine Learning (ML) compiler, ML inference runtime and software/hardware co-design for high-concurrency ML serving systems. During my PhD, I spent most of my time in the area of electronic system-level (ESL) design and modeling, mainly focusing on distributed runtime/middleware and system-level performance modeling for edge computing systems.

Phone: +1-512-751-1819
E-mail: zhuoran [@] utexas [DOT] edu

Distributed Adaptive Deep Learning Inference Framework (DeepThings): DeepThings is a framework for locally distributed and adaptive CNN inference in resource-constrained IoT edge clusters, which mainly consists of:

• A Fused Tile Partitioning (FTP) method for dividing convolutional layers into independently distributable tasks. FTP fuses layers and partitions them vertically in a grid fashion, which largely reduces communication and task migration overhead.

• A distributed work stealing runtime system for IoT clusters to adaptively distribute FTP partitions in dynamic application scenarios.

Z. Zhao, K. Mirzazad and A. Gerstlauer, "DeepThings: Distributed Adaptive Deep Learning Inference on Resource-Constrained IoT Edge Clusters," CODES+ISSS, special issue of IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems (TCAD), 2018.

[PDF] [Code] [Slides] [Poster]

Network-of-Systems Simulator (NoSSim): A source-level network/system co-simulation framework for rapid embedded/mobile system prototyping, which combines:

• A Host-Compiled SystemC full system simulation platform with an abstract OS model.

• A LLVM-based function-level profiling and instrumentation tool.

• Network interaction emulation based on OMNeT++ network simulation framework and lwIP library.

Z. Zhao, V. Tsoutsouras, D. Soudris and A. Gerstlauer, "Network/System Co-Simulation for Design Space Exploration of IoT Applications," in Proceedings of the International Conference on Embedded Computer Systems: Architectures, Modeling and Simulation (SAMOS), 2017.

[PDF][Code] [Slides]

Retargetable Back-Annotator (RBA): A compile-time profiling and instrumentation tool for source-level system performance evaluation.

Z. Zhao, A. Gerstlauer and L. K. John, "Source-Level Performance, Energy, Reliability, Power and Thermal (PERPT) Simulation," in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems (TCAD), 2017.

[PDF][Code] [Slides] [Poster]