Summary
Sicong Zhuang is a Senior ML Kernel Engineer with 12 years of experience specializing in deep learning kernels, compiler backends, and high-performance computing for next-gen hardware. He builds and validates high-performance kernels for GPT3-like models, optimizes DL compiler backends (Facebook Glow, Apache TVM), and leverages mixed-precision numerical techniques to squeeze performance from novel architectures at Cerebras. His background includes parallel BLAS/LAPACK development, sparse numerical methods and a cycle-accurate CPU simulator from his PhD and HPC work at Barcelona Supercomputing Center, giving him rare cross-domain expertise in both algorithms and microarchitecture. Sicong also publishes open-source tooling for distributed training (bitpack, altsplit) that reduce communication and memory overhead in multi-GPU/MPI setups, reflecting a pragmatic focus on production-scale efficiency. Based in Toronto, he combines research rigor with hands-on hardware bring-up and low-level debugging to translate architectural features into reliable ML performance.
12 years of coding experience
9 years of employment as a software developer
Bachelor's degree, Management Information Systems, Bachelor's degree, Management Information Systems at University of Tasmania
UPC Universitat Politècnica de Catalunya
English, Chinese, Spanish, German, French