Zihang Chen

Currently pursuing M.S. degree in Computer Science and Technology. Familiarity with popular CNN models, lightweight models, and binary CNNs. Comprehensive understanding of computer architecture. Experience in DL training with CUDA GPU and inference acceleration with FPGAs. Familiar with parallel programming using CUDA/OpenMP/MPI.

Second prize in operating system (OS) kernel competition 2022 held by Computer Education Research Association of Chinese Universities (CERCU).

The kernel is written using mainly C/C++ together with RISC-V assembly. The hardware platform is Kendryte k210 SoC featuring a dual-core RISC-V processor. Our design includes system boot/initialization, process management, memory management including paging/virtualization, and file system. The project demonstrates my proficiency in C/C++ programming, familiarity with related toolchains as well as a deep understanding of computing architecture. The code is publicly available at: https://gitlab.eduxiji.net/educg-group-14238-914330/CG2019011412-1625

Satellite based AI computing platform and lightweight CNNs.

Cooperation with an aerospace institute in developing AI computing platform capable of satellite-level and component-level detection and pose estimation based on various convolutional NN models.

I took responsibly in writing C/C++ inference code for ground-truth comparison as well as Verilog HDL code for each hardware modules including convolution, residual addition, pooling, etc.

Furthermore, my current research interest is lightweight NNs particularly binary neural networks (BNNs). I have experience in training state-of-the-art BNN models such as ReActNet and Bi-Real Net, using PyTorch framework.

Parallel computing

Got familiar with parallel programming including CUDA, OpenMP and MPI during undergraduate classes.