Yang Chen

My current focus is on leveraging advanced compiler techniques to accelerate deep neural networks (DNNs). I have extensive experience working with DNN compiler toolchains such as PyTorch, AITemplate, Triton, MLIR, ONNXRuntime and TVM. My work ranges from improving the performance of model inference for CPUs to generating high performant kernels for NVIDIA GPUs. I also contribute to leading inference platforms like vLLM and implement optimized CUDA, CUTLASS, and Triton kernels to push the boundaries of efficiency and scalability. In the past, I had years of experience with a wide spectrum of the "traditional" compiler stack, including compiler front-ends, program transformations, superoptimization, and compiler testing. * I am humbled by receiving a Google Open Source Peer Bonus for my contribution to C-Reduce: https://opensource.googleblog.com/2018/03/congratulating-open-source-peer-bonus-winners.html * My GitHub page: https://github.com/chenyang78 * My Google Scholar page: https://scholar.google.com/citations?user=dgFuNeEAAAAJ&hl=en * Academic Awards: The ACM SIGPLAN Most Influential PLDI 2012 Paper Award for "Test-case reduction for C compiler bugs" The ACM SIGPLAN Most Influential PLDI 2011 Paper Award for "Finding and Understanding Bugs in C Compilers"